From david.darr at gmail.com Tue Jul 7 14:29:28 2009 From: david.darr at gmail.com (david darr) Date: Tue, 7 Jul 2009 14:29:28 -0700 Subject: [Warewulf] different passwd files for different nodes Message-ID: <6a8af7e80907071429n35b1ba09n2e6a59eeb8b3aeac@mail.gmail.com> Hello all, This is probably a very simple question but I haven't figured it out yet so thought I would ask. I am currently running the "passwdfile module" for all of my nodes and the passwd file that is distributed to my nodes is /etc/perceus/modules/passwdfile/all. I now want to divide my nodes into groups that use different passwd files. How can I do this? I am pretty sure that I could also do this, in a slightly different way, by using SLURM, but for a variety of reasons I don't want to go that path (at least not yet). thannks, David -- -------------------------------------------------------------------------- David Darr - ph: (206) 616-4953 - email: darr at ocean.washington.edu School of Oceanography - Univ. of Washington - Seattle, WA. 98195 -------------------------------------------------------------------------- -------------- next part -------------- An HTML attachment was scrubbed... URL: http://altruistic.infiscale.org/pipermail/perceus/attachments/20090707/731382d7/attachment.html From benedict.verhegghe at ugent.be Thu Jul 9 03:47:33 2009 From: benedict.verhegghe at ugent.be (Benedict Verhegghe) Date: Thu, 09 Jul 2009 12:47:33 +0200 Subject: [Warewulf] Problem registering perceus In-Reply-To: References: Message-ID: <4A55CAC5.2080503@ugent.be> I think that the node not coming up is not related with not registering. Registration always fails for me too, and I have three clusters running on perceus. The bad thing is that the perceus command keeps insisting on regsitration and then hangs for a long time until something times out. My solution has always been to comment out the following lines in scripts/lib/Perceus/Configure.pm: # if ( ! &contact_register() ) { # &getinput("Press [ENTER] to continue: "); # } That will make perceus skip the registration. And like Greg asks in the comments in that file: please do not use this just to avoid registration, only temporarily to get perceus working. I always tought the problem was with my custom installation (Debian, and a patched Perceus), but since you're having the problems on Caos, the developers might have a look at a possible fix or work-around. Maybe just have perceus write a message to 'Please register' and have a 'perceus register' command do the actual registration? Benedict Woon-Seng Choong schreef: > I am trying to install Caos NSA and perceus. Let me just say that I have no problem registering with perceus, but I keep getting error registering perceus. I am trying to setup a cluster for a collaborator in Malaysia, and initially I don't even connect to the outside world. Every time I try to run a perceus command, it jumps into the registration. And when I try to boot up a node, it doesn't come up completely and I don't know if this is due to not registering. Is there a way to bypass the registration and make the node boot up properly. > > Thank you > Seng > > > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf From angelv at iac.es Thu Jul 9 06:04:35 2009 From: angelv at iac.es (angelv at iac.es) Date: Thu, 09 Jul 2009 14:04:35 +0100 Subject: [Warewulf] Initial problems with Caos NSA - Perceus Message-ID: <20090709140435.yngte6uycgcgksc8@vangogh.ll.iac.es> Hi, after years of having Warewulf - Perceus in the radar, I have decided now to have a try at it, so I got a small 4-node cluster to try it on. I downloaded Caos NSA 1.0.8 for 64 bits, and installed in the head node. The installation seemed to go fine, except for a few problems with the keyboard, but I had no luck with the Clustering configuration. I was initially following http://www.linux-mag.com/id/7239, but apparently some things have changed since that was published, so instead I followed http://wiki.caoslinux.org/Caos_NSA_1_Installation_Walkthrough for CAOS installation and then http://wiki.caoslinux.org/Clustering for the clustering part. But things didn't go smoothly... When configuring Perceus (as part of the first installation process, before the first reboot), it asks me to get a VNFS capsule. I say yes for the default one. It gets a "stateless" capsule and automatically configures eth0 with a 10.x.x.x address and says it is ready for 500 nodes. Then after rebooting I turn on one of the salve nodes and check the /var/log/messages. It does recognize that someone is attempting to boot, it gives it a temporary address:10.254.3.243, but it failes when trying to send the pxelinux.0 At the same time, it seems that the DHCPD server is not configured to hand DHCP addresses only to the cluster network. Then I try to follow the instructions in http://wiki.caoslinux.org/Clustering. I open sidekick and try to configure the section Clustering, but it does not seem to do the correct thing. I configure it with address 10.254.0.1-4, but when finished /etc/hosts has now many many IPs in the 10.254.x.x subnets and I still get the sampe problems as above. Any idea what could be wrong before I start looking under the hood? Thanks a lot, ?ngel de Vicente From jason at redbarncomputers.com Thu Jul 9 06:39:20 2009 From: jason at redbarncomputers.com (Jason Somers) Date: Thu, 09 Jul 2009 09:39:20 -0400 Subject: [Warewulf] Initial problems with Caos NSA - Perceus In-Reply-To: <20090709140435.yngte6uycgcgksc8@vangogh.ll.iac.es> References: <20090709140435.yngte6uycgcgksc8@vangogh.ll.iac.es> Message-ID: <4A55F308.2010200@redbarncomputers.com> Angel, We experienced the same problem recently on nodes with newer Intel GB NICs. (older NICs worked fine) After updating Perceus from 1.5.0 (that comes stock in NSA 1.0.8) to 1.5.2, the problem went away... # smart upgrade perceus* Give it a shot, and let us know. -Jason ============================ Jason Somers Network Administrator Red Barn Computers 1235 Front Street - Suite #3 Binghamton, NY 13905 (607) 772-1888 x222 angelv at iac.es wrote: > Hi, > > after years of having Warewulf - Perceus in the radar, I have decided > now to have a try at it, so I got a small 4-node cluster to try it on. > > I downloaded Caos NSA 1.0.8 for 64 bits, and installed in the head > node. The installation seemed to go fine, except for a few problems > with the keyboard, but I had no luck with the Clustering > configuration. I was initially following > http://www.linux-mag.com/id/7239, but apparently some things have > changed since that was published, so instead I followed > http://wiki.caoslinux.org/Caos_NSA_1_Installation_Walkthrough for CAOS > installation and then http://wiki.caoslinux.org/Clustering for the > clustering part. > > But things didn't go smoothly... > > When configuring Perceus (as part of the first installation process, > before the first reboot), it asks me to get a VNFS capsule. I say yes > for the default one. It gets a "stateless" capsule and automatically > configures eth0 with a 10.x.x.x address and says it is ready for 500 > nodes. Then after rebooting I turn on one of the salve nodes and check > the /var/log/messages. It does recognize that someone is attempting to > boot, it gives it a temporary address:10.254.3.243, but it failes when > trying to send the pxelinux.0 At the same time, it seems that the > DHCPD server is not configured to hand DHCP addresses only to the > cluster network. > > Then I try to follow the instructions in > http://wiki.caoslinux.org/Clustering. I open sidekick and try to > configure the section Clustering, but it does not seem to do the > correct thing. I configure it with address 10.254.0.1-4, but when > finished /etc/hosts has now many many IPs in the 10.254.x.x subnets > and I still get the sampe problems as above. > > Any idea what could be wrong before I start looking under the hood? > > Thanks a lot, > ?ngel de Vicente > > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf > From angelv at iac.es Thu Jul 9 07:11:59 2009 From: angelv at iac.es (angelv at iac.es) Date: Thu, 09 Jul 2009 15:11:59 +0100 Subject: [Warewulf] Initial problems with Caos NSA - Perceus Message-ID: <20090709151159.dokx6dfkxwogg8ws@vangogh.ll.iac.es> Hi, ========================= Angel, We experienced the same problem recently on nodes with newer Intel GB NICs. (older NICs worked fine) After updating Perceus from 1.5.0 (that comes stock in NSA 1.0.8) to 1.5.2, the problem went away... # smart upgrade perceus* Give it a shot, and let us know. -Jason ========================= Thanks for the advice. I tried, but still get the same problem. Perhaps my problem is the opposite, that my cards are too old? After attaching a monitor to the slave node, I see that it refuses to recognize the card and aborts with the following message: ERROR: No supported network interace cards found! The NIC (as reported by Caos in the head node, all cards in all the nodes are the same) is Tigon3 [partno(BCM95704A7) rev 2003 PHY(5704)] Thanks a lot, ?ngel de Vicente From gmkurtzer at gmail.com Thu Jul 9 08:14:21 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Thu, 9 Jul 2009 08:14:21 -0700 Subject: [Warewulf] Initial problems with Caos NSA - Perceus In-Reply-To: <20090709151159.dokx6dfkxwogg8ws@vangogh.ll.iac.es> References: <20090709151159.dokx6dfkxwogg8ws@vangogh.ll.iac.es> Message-ID: <571f1a060907090814u7f344a28m896f3332e6563104@mail.gmail.com> Hello, I think you might be running into a problem that we have already fixed in the Perceus 1.6.0 that we are testing now. If you would like to test the 1.6.0 kernel and initramfs, grab the following tarball and decompress it and replace that kernel and initramfs.img for the existing ones (depending on your build in /var/lib/perceus/tftp). Of course back up your original ones as well. http://www.perceus.org/gmk/perceus-1.6.0-kernel.tar.gz Let me know how that works for you. Thanks, Greg On Thu, Jul 9, 2009 at 7:11 AM, wrote: > Hi, > > ========================= > Angel, > ? ? We experienced the same problem recently on nodes with newer Intel > GB NICs. (older NICs worked fine) After updating Perceus from 1.5.0 > (that comes stock in NSA 1.0.8) to 1.5.2, the problem went away... > > # smart upgrade perceus* > > Give it a shot, and let us know. > -Jason > ========================= > > Thanks for the advice. I tried, but still get the same problem. > Perhaps my problem is the opposite, that my cards are too old? After > attaching a monitor to the slave node, I see that it refuses to > recognize the card and aborts with the following message: > > ERROR: No supported network interace cards found! > > The NIC (as reported by Caos in the head node, all cards in all the > nodes are the same) is Tigon3 [partno(BCM95704A7) rev 2003 PHY(5704)] > > Thanks a lot, > ?ngel de Vicente > > > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From gmkurtzer at gmail.com Thu Jul 9 08:20:35 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Thu, 9 Jul 2009 08:20:35 -0700 Subject: [Warewulf] Initial problems with Caos NSA - Perceus In-Reply-To: <4A55F308.2010200@redbarncomputers.com> References: <20090709140435.yngte6uycgcgksc8@vangogh.ll.iac.es> <4A55F308.2010200@redbarncomputers.com> Message-ID: <571f1a060907090820s47b0a608o42e27d83dc78a96c@mail.gmail.com> Hello, Yes as Jason mentioned an upgrade might be in order (see my previous email) but with that said I think you have a bit more going on as well. There are normally 2 requests to get the PXE files. The first one is a query to make sure the file is there which shows up as a failure to send in the logs. The following should work. Is that what you are seeing as well? If the previous doesn't work, please let us know how far exactly the node is getting and we can help some more. Greg On Thu, Jul 9, 2009 at 6:39 AM, Jason Somers wrote: > Angel, > ? ?We experienced the same problem recently on nodes with newer Intel > GB NICs. (older NICs worked fine) After updating Perceus from 1.5.0 > (that comes stock in NSA 1.0.8) to 1.5.2, the problem went away... > > # smart upgrade perceus* > > Give it a shot, and let us know. > -Jason > > ============================ > Jason Somers > Network Administrator > Red Barn Computers > 1235 Front Street - Suite #3 > Binghamton, NY 13905 > (607) 772-1888 x222 > > > > angelv at iac.es wrote: >> Hi, >> >> after years of having Warewulf - Perceus in the radar, I have decided >> now to have a try at it, so I got a small 4-node cluster to try it on. >> >> I downloaded Caos NSA 1.0.8 for 64 bits, and installed in the head >> node. The installation seemed to go fine, except for a few problems >> with the keyboard, but I had no luck with the Clustering >> configuration. I was initially following >> http://www.linux-mag.com/id/7239, but apparently some things have >> changed since that was published, so instead I followed >> http://wiki.caoslinux.org/Caos_NSA_1_Installation_Walkthrough for CAOS >> installation and then http://wiki.caoslinux.org/Clustering for the >> clustering part. >> >> But things didn't go smoothly... >> >> When configuring Perceus (as part of the first installation process, >> before the first reboot), it asks me to get a VNFS capsule. I say yes >> for the default one. It gets a "stateless" capsule and automatically >> configures eth0 with a 10.x.x.x ?address and says it is ready for 500 >> nodes. Then after rebooting I turn on one of the salve nodes and check >> the /var/log/messages. It does recognize that someone is attempting to >> boot, it gives it a temporary address:10.254.3.243, but it failes when >> trying to send the pxelinux.0 At the same time, it seems that the >> DHCPD server is not configured to hand DHCP addresses only to the >> cluster network. >> >> Then I try to follow the instructions in >> http://wiki.caoslinux.org/Clustering. I open sidekick and try to >> configure the section Clustering, but it does not seem to do the >> correct thing. I configure it with address 10.254.0.1-4, but when >> finished /etc/hosts has now many many IPs in the 10.254.x.x subnets >> and I still get the sampe problems as above. >> >> Any idea what could be wrong before I start looking under the hood? >> >> Thanks a lot, >> ?ngel de Vicente >> >> _______________________________________________ >> Warewulf mailing list >> Warewulf at caoslinux.org >> http://lists.caosity.org/mailman/listinfo/warewulf >> > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From gmkurtzer at gmail.com Thu Jul 9 08:31:00 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Thu, 9 Jul 2009 08:31:00 -0700 Subject: [Warewulf] Problem registering perceus In-Reply-To: <4A55CAC5.2080503@ugent.be> References: <4A55CAC5.2080503@ugent.be> Message-ID: <571f1a060907090831p26b98f4dy540b256045bfa71a@mail.gmail.com> Hello, There were some changes in how registration got done from 1.5.1 to 1.5.2 so chances are an upgrade would cause a second registration. The registration is obviously not suppose to interfere with normal usage or be a pain to deal with. It is only there so we as developers of the software can see who is installing it and we use it for statistics for presentations, and periodically we contact the people registering asking if they need assistance based on their comments or repetitive registrations. We do read every registration and we do enjoy getting them, but as I mentioned we do not want it interfering or annoying people. So if anything isn't working properly, it needs to get fixed. In the end, this isn't closed source, and commenting it out in the source code can work around the problem. With that said, I would like to solve the problem, so if anyone is still seeing the registration problem please let us know with the version of the Perceus and what exactly happens after you finished typing in your registration information. Thank you! Greg On Thu, Jul 9, 2009 at 3:47 AM, Benedict Verhegghe wrote: > I think that the node not coming up is not related with not registering. > Registration always fails for me too, and I have three clusters running > on perceus. > > The bad thing is that the perceus command keeps insisting on > regsitration and then hangs for a long time until something times out. > My solution has always been to comment out the following lines in > scripts/lib/Perceus/Configure.pm: > > # ? ? ? ? if ( ! &contact_register() ) { > # ? ? ? ? ? ?&getinput("Press [ENTER] to continue: "); > # ? ? ? ? } > > That will make perceus skip the registration. > And like Greg asks in the comments in that file: please do not use this > just to avoid registration, only temporarily to get perceus working. > > I always tought the problem was with my custom installation (Debian, and > a patched Perceus), but since you're having the problems on Caos, the > developers might have a look at a possible fix or work-around. > Maybe just have perceus write a message to 'Please register' and have > a 'perceus register' command do the actual registration? > > Benedict > > > Woon-Seng Choong schreef: >> I am trying to install Caos NSA and perceus. Let me just say that I have no problem registering with perceus, but I keep getting error registering perceus. I am trying to setup a cluster for a collaborator in Malaysia, and initially I don't even connect to the outside world. Every time I try to run a perceus command, it jumps into the registration. And when I try to boot up a node, it doesn't come up completely and I don't know if this is due to not registering. Is there a way to bypass the registration and make the node boot up properly. >> >> ? ? ?Thank you >> ? ? ?Seng >> >> >> _______________________________________________ >> Warewulf mailing list >> Warewulf at caoslinux.org >> http://lists.caosity.org/mailman/listinfo/warewulf > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From angelv at iac.es Fri Jul 10 01:44:44 2009 From: angelv at iac.es (=?ISO-8859-1?Q?=C1ngel_de_Vicente?=) Date: Fri, 10 Jul 2009 09:44:44 +0100 Subject: [Warewulf] Initial problems with Caos NSA - Perceus In-Reply-To: <20090709151159.dokx6dfkxwogg8ws@vangogh.ll.iac.es> References: <20090709151159.dokx6dfkxwogg8ws@vangogh.ll.iac.es> Message-ID: <4A56FF7C.8000604@iac.es> Hi, thanks for the help, but I'm afraid substituting the kernel and the initramfs.img was not enough... In the master node messages file I get: Jul 10 09:21:43 localhost perceus-dnsmasq[2429]: DHCPDISCOVER(eth0) 00:e0:81:33:20:46 Jul 10 09:21:43 localhost perceus-dnsmasq[2429]: DHCPOFFER(eth0) 10.254.3.243 00:e0:81:33:20:46 Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: DHCPREQUEST(eth0) 10.254.3.243 00:e0:81:33:20:46 Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: DHCPACK(eth0) 10.254.3.243 00:e0:81:33:20:46 Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/pxelinux.0 to 10.254.3.243 Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: TFTP error 0 TFTP Aborted received from 10.254.3.243 Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: TFTP failed sending /var/lib/perceus/tftp/pxelinux.0 to 10.254.3.243 Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/pxelinux.0 to 10.254.3.243 Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/pxelinux.cfg/default to 10.254.3.243 Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/vesamenu.c32 to 10.254.3.243 Jul 10 09:21:48 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/pxelinux.cfg/default to 10.254.3.243 Jul 10 09:21:48 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/perceus-splash.png to 10.254.3.243 Jul 10 09:22:21 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/kernel to 10.254.3.243 Jul 10 09:22:21 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/initramfs.img to 10.254.3.243 So apparently it goes through the proper motions, as you mentioned in a previous post, first giving the error message and then sending it properly. Everything seems to be sent to the slave node properly, but after showing the splash screen the slave node comes up with the same error message as before: /ERROR: No supported network interace cards found! / I guess it is a matter of including the appropriate drivers in the kernel? It is a long time since I do anything like that, so any help on how to proceed would be welcome. Thanks, ?ngel de Vicente ============================= Hello, I think you might be running into a problem that we have already fixed in the Perceus 1.6.0 that we are testing now. If you would like to test the 1.6.0 kernel and initramfs, grab the following tarball and decompress it and replace that kernel and initramfs.img for the existing ones (depending on your build in /var/lib/perceus/tftp). Of course back up your original ones as well. http://www.perceus.org/gmk/perceus-1.6.0-kernel.tar.gz Let me know how that works for you. Thanks, Greg ============================ --------------------------------------------------------------------------------------------- ADVERTENCIA: Sobre la privacidad y cumplimiento de la Ley de Protecci?n de Datos, acceda a http://www.iac.es/disclaimer.php WARNING: For more information on privacy and fulfilment of the Law concerning the Protection of Data, consult http://www.iac.es/disclaimer.php?lang=en From astevens at Infiscale.com Fri Jul 10 06:48:13 2009 From: astevens at Infiscale.com (Arthur A. Stevens) Date: Fri, 10 Jul 2009 06:48:13 -0700 Subject: [Warewulf] Initial problems with Caos NSA - Perceus In-Reply-To: <4A56FF7C.8000604@iac.es> References: <20090709151159.dokx6dfkxwogg8ws@vangogh.ll.iac.es> <4A56FF7C.8000604@iac.es> Message-ID: <4A57469D.6060001@Infiscale.com> Can you send in as much information on that system as you can? Add in cards, bios version, even who made the case if possible. Thanks. Arthur ?ngel de Vicente wrote: > Hi, > > thanks for the help, but I'm afraid substituting the kernel and the initramfs.img was not enough... > > In the master node messages file I get: > > Jul 10 09:21:43 localhost perceus-dnsmasq[2429]: DHCPDISCOVER(eth0) 00:e0:81:33:20:46 > Jul 10 09:21:43 localhost perceus-dnsmasq[2429]: DHCPOFFER(eth0) 10.254.3.243 00:e0:81:33:20:46 > Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: DHCPREQUEST(eth0) 10.254.3.243 00:e0:81:33:20:46 > Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: DHCPACK(eth0) 10.254.3.243 00:e0:81:33:20:46 > Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/pxelinux.0 to 10.254.3.243 > Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: TFTP error 0 TFTP Aborted received from 10.254.3.243 > Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: TFTP failed sending /var/lib/perceus/tftp/pxelinux.0 to 10.254.3.243 > Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/pxelinux.0 to 10.254.3.243 > Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/pxelinux.cfg/default to 10.254.3.243 > Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/vesamenu.c32 to 10.254.3.243 > Jul 10 09:21:48 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/pxelinux.cfg/default to 10.254.3.243 > Jul 10 09:21:48 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/perceus-splash.png to 10.254.3.243 > Jul 10 09:22:21 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/kernel to 10.254.3.243 > Jul 10 09:22:21 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/initramfs.img to 10.254.3.243 > > So apparently it goes through the proper motions, as you mentioned in a previous post, first giving the error message and then sending it properly. > Everything seems to be sent to the slave node properly, but after showing the splash screen the slave node comes up with the same error message as before: > > /ERROR: No supported network interace cards found! > / > I guess it is a matter of including the appropriate drivers in the kernel? It is a long time since I do anything like that, so any help on how to proceed would be welcome. > > Thanks, > ?ngel de Vicente > ============================= > > Hello, > > I think you might be running into a problem that we have already fixed > in the Perceus 1.6.0 that we are testing now. > > If you would like to test the 1.6.0 kernel and initramfs, grab the > following tarball and decompress it and replace that kernel and > initramfs.img for the existing ones (depending on your build in > /var/lib/perceus/tftp). Of course back up your original ones as well. > > http://www.perceus.org/gmk/perceus-1.6.0-kernel.tar.gz > > Let me know how that works for you. > > Thanks, > Greg > ============================ > > --------------------------------------------------------------------------------------------- > ADVERTENCIA: Sobre la privacidad y cumplimiento de la Ley de Protecci?n de Datos, acceda a http://www.iac.es/disclaimer.php > WARNING: For more information on privacy and fulfilment of the Law concerning the Protection of Data, consult http://www.iac.es/disclaimer.php?lang=en > > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf > From gmkurtzer at gmail.com Fri Jul 10 11:15:03 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Fri, 10 Jul 2009 11:15:03 -0700 Subject: [Warewulf] Initial problems with Caos NSA - Perceus In-Reply-To: <4A57469D.6060001@Infiscale.com> References: <20090709151159.dokx6dfkxwogg8ws@vangogh.ll.iac.es> <4A56FF7C.8000604@iac.es> <4A57469D.6060001@Infiscale.com> Message-ID: <571f1a060907101115s693c33ebwb689f20f0ceb3c11@mail.gmail.com> And in addition to what Arthur mentioned, the output of: # lspci; lspci -n on a system with the same network interface would be very helpful in resolving this. Thank you. Greg On Fri, Jul 10, 2009 at 6:48 AM, Arthur A. Stevens wrote: > Can you send in as much information on that system as you can? Add in > cards, bios version, even who made the case if possible. Thanks. > > Arthur > > ?ngel de Vicente wrote: >> Hi, >> >> thanks for the help, but I'm afraid substituting the kernel and the initramfs.img was not enough... >> >> In the master node messages file I get: >> >> Jul 10 09:21:43 localhost perceus-dnsmasq[2429]: DHCPDISCOVER(eth0) 00:e0:81:33:20:46 >> Jul 10 09:21:43 localhost perceus-dnsmasq[2429]: DHCPOFFER(eth0) 10.254.3.243 00:e0:81:33:20:46 >> Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: DHCPREQUEST(eth0) 10.254.3.243 00:e0:81:33:20:46 >> Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: DHCPACK(eth0) 10.254.3.243 00:e0:81:33:20:46 >> Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/pxelinux.0 to 10.254.3.243 >> Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: TFTP error 0 TFTP Aborted received from 10.254.3.243 >> Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: TFTP failed sending /var/lib/perceus/tftp/pxelinux.0 to 10.254.3.243 >> Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/pxelinux.0 to 10.254.3.243 >> Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/pxelinux.cfg/default to 10.254.3.243 >> Jul 10 09:21:47 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/vesamenu.c32 to 10.254.3.243 >> Jul 10 09:21:48 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/pxelinux.cfg/default to 10.254.3.243 >> Jul 10 09:21:48 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/perceus-splash.png to 10.254.3.243 >> Jul 10 09:22:21 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/kernel to 10.254.3.243 >> Jul 10 09:22:21 localhost perceus-dnsmasq[2429]: TFTP sent /var/lib/perceus/tftp/initramfs.img to 10.254.3.243 >> >> So apparently it goes through the proper motions, as you mentioned in a previous post, first giving the error message and then sending it properly. >> Everything seems to be sent to the slave node properly, but after showing the splash screen the slave node comes up with the same error message as before: >> >> /ERROR: No supported network interace cards found! >> / >> I guess it is a matter of including the appropriate drivers in the kernel? It is a long time since I do anything like that, so any help on how to proceed would be welcome. >> >> Thanks, >> ?ngel de Vicente >> ============================= >> >> Hello, >> >> I think you might be running into a problem that we have already fixed >> in the Perceus 1.6.0 that we are testing now. >> >> If you would like to test the 1.6.0 kernel and initramfs, grab the >> following tarball and decompress it and replace that kernel and >> initramfs.img for the existing ones (depending on your build in >> /var/lib/perceus/tftp). Of course back up your original ones as well. >> >> http://www.perceus.org/gmk/perceus-1.6.0-kernel.tar.gz >> >> Let me know how that works for you. >> >> Thanks, >> Greg >> ============================ >> >> --------------------------------------------------------------------------------------------- >> ADVERTENCIA: Sobre la privacidad y cumplimiento de la Ley de Protecci?n de Datos, acceda a http://www.iac.es/disclaimer.php >> WARNING: For more information on privacy and fulfilment of the Law concerning the Protection of Data, consult http://www.iac.es/disclaimer.php?lang=en >> >> _______________________________________________ >> Warewulf mailing list >> Warewulf at caoslinux.org >> http://lists.caosity.org/mailman/listinfo/warewulf >> > > > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From bernard at vanhpc.org Mon Jul 13 12:05:39 2009 From: bernard at vanhpc.org (Bernard Li) Date: Mon, 13 Jul 2009 12:05:39 -0700 Subject: [Warewulf] Turning off hybridization in Perceus Message-ID: Hi guys: Currently, when configuring a VNFS image (either by chroot2stateless.sh or by perceus vnfs configure), it would ask whether you want to enable hybridization or not. If you answer no, it would simply not set up fstab entry for the statedir. The issue is that the hybridize file is still created in your VNFS image dir and the default directories are still "hybridized", so the end result is that a node when booted against that image, would have dangling symlinks to these directories. The current workaround is simply to uncomment the directories in the hybridize file, but it would be nice if there was a way to simply turn hybridization off (maybe through a config option?) Thanks, Bernard From bernard at vanhpc.org Mon Jul 13 14:46:06 2009 From: bernard at vanhpc.org (Bernard Li) Date: Mon, 13 Jul 2009 14:46:06 -0700 Subject: [Warewulf] [PATCH] Perceus 1.6 Configure.pm missing semi-colons Message-ID: Hi guys: Just checked out 1.6 branch to give it a spin, and found that Configure.pm is missing semi-colon for multiple lines, the attached patch fixes that. Cheers, Bernard -------------- next part -------------- A non-text attachment was scrubbed... Name: Configure.pm.patch Type: text/x-patch Size: 9157 bytes Desc: not available Url : http://altruistic.infiscale.org/pipermail/perceus/attachments/20090713/28a9fe17/attachment.bin From bernard at vanhpc.org Thu Jul 16 11:33:03 2009 From: bernard at vanhpc.org (Bernard Li) Date: Thu, 16 Jul 2009 11:33:03 -0700 Subject: [Warewulf] Bug with yum and chroot Message-ID: Hi all: If you are using yum 3.2.18+ and have tried to install yum packages via --installroot into a VNFS image, for instance: yum --installroot /var/lib/perceus/vnfs//rootfs update blah Then you may notice some transaction-* logs being placed in /var/lib/perceus/vnfs//rootfs/var/lib/perceus/vnfs//rootfs/var/lib/yum. This is a bug, they should actually be placed in /var/lib/perceus/vnfs//rootfs/var/lib/yum (in fact, they should be transient unless something went wrong during the install). A bug has already been filed, and you can track the progress here: https://bugzilla.redhat.com/show_bug.cgi?id=503195 As far as I can tell, this bug is not fatal, but just a bit of an annoyance. Cheers, Bernard From gwleong at gmail.com Thu Jul 16 19:15:28 2009 From: gwleong at gmail.com (Gary Leong) Date: Thu, 16 Jul 2009 19:15:28 -0700 Subject: [Warewulf] links on vnfs capsules Message-ID: I'm not using the hybridization model, but I'm getting these links when I boot up. Hmm...where do the links come from? When I do a vnfs mount, I don't see the links, but libraries I need. But when I boot up, something blows away my /usr/share /usr/src and /usr/include and replaces it with this links that are blinking because they don't connect to anywhere. A lot of things break because of this. Any help would be appreciated. Thanks. gary lrwxrwxrwx 1 root root 59 Jun 28 06:00 include -> /var/lib/perceus//vnfs/Centos-5.3.x86_64/rootfs/usr/include drwxr-xr-x 4 root root 0 Jun 28 06:00 kerberos drwxr-xr-x 13 root root 0 Jun 28 06:00 lib drwxr-xr-x 19 root root 0 Jun 28 06:00 lib64 drwxr-xr-x 5 root root 0 Jun 28 06:00 libexec drwxr-xr-x 12 root root 0 Jun 28 06:00 local drwxr-xr-x 3 root root 0 Jun 28 06:00 man drwxr-xr-x 2 root root 0 Jun 28 06:00 sbin lrwxrwxrwx 1 root root 57 Jun 28 06:00 share -> /var/lib/perceus//vnfs/Centos-5.3.x86_64/rootfs/usr/share lrwxrwxrwx 1 root root 55 Jun 28 06:00 src -> /var/lib/perceus//vnfs/Centos-5.3.x86_64/rootfs/usr/src lrwxrwxrwx 1 root root 10 Jun 28 06:00 tmp -> ../var/tmp -------------- next part -------------- An HTML attachment was scrubbed... URL: http://altruistic.infiscale.org/pipermail/perceus/attachments/20090716/a8684e7d/attachment.html From gwleong at gmail.com Thu Jul 16 19:17:51 2009 From: gwleong at gmail.com (Gary Leong) Date: Thu, 16 Jul 2009 19:17:51 -0700 Subject: [Warewulf] links on vnfs capsules In-Reply-To: References: Message-ID: Actually, I see when I do a perceus vnfs mount capsule It happens really quick, to quick. It's not uncompressing my capsule. It's linking to my a copy of chroot. Therefore, that's why the links aren't showing, ah? I have to dismantle the capsule myself to see the links, ah? Gary On Thu, Jul 16, 2009 at 7:15 PM, Gary Leong wrote: > I'm not using the hybridization model, but I'm getting these links when I > boot up. Hmm...where do the links come from? When I do a vnfs mount, I > don't see the links, but libraries I need. But when I boot up, something > blows away my /usr/share /usr/src and /usr/include and replaces it with this > links that are blinking because they don't connect to anywhere. A lot of > things break because of this. > > Any help would be appreciated. Thanks. > > gary > > lrwxrwxrwx 1 root root 59 Jun 28 06:00 include -> > /var/lib/perceus//vnfs/Centos-5.3.x86_64/rootfs/usr/include > drwxr-xr-x 4 root root 0 Jun 28 06:00 kerberos > drwxr-xr-x 13 root root 0 Jun 28 06:00 lib > drwxr-xr-x 19 root root 0 Jun 28 06:00 lib64 > drwxr-xr-x 5 root root 0 Jun 28 06:00 libexec > drwxr-xr-x 12 root root 0 Jun 28 06:00 local > drwxr-xr-x 3 root root 0 Jun 28 06:00 man > drwxr-xr-x 2 root root 0 Jun 28 06:00 sbin > lrwxrwxrwx 1 root root 57 Jun 28 06:00 share -> > /var/lib/perceus//vnfs/Centos-5.3.x86_64/rootfs/usr/share > lrwxrwxrwx 1 root root 55 Jun 28 06:00 src -> > /var/lib/perceus//vnfs/Centos-5.3.x86_64/rootfs/usr/src > lrwxrwxrwx 1 root root 10 Jun 28 06:00 tmp -> ../var/tmp > > -------------- next part -------------- An HTML attachment was scrubbed... URL: http://altruistic.infiscale.org/pipermail/perceus/attachments/20090716/82d6e8d0/attachment.html From angelv at iac.es Fri Jul 17 01:11:46 2009 From: angelv at iac.es (=?ISO-8859-1?Q?=C1ngel_de_Vicente?=) Date: Fri, 17 Jul 2009 09:11:46 +0100 Subject: [Warewulf] Initial problems with Caos NSA - Perceus In-Reply-To: References: Message-ID: <4A603242.1010706@iac.es> Hi, sorry for the late reply to this... It was a VERY busy week... > ---------------------------------------------------------------------- > > Message: 1 > Date: Fri, 10 Jul 2009 06:48:13 -0700 > From: "Arthur A. Stevens" > Subject: Re: [Warewulf] Initial problems with Caos NSA - Perceus > To: The Warewulf Cluster Toolkit > Message-ID: <4A57469D.6060001 at Infiscale.com> > Content-Type: text/plain; charset=ISO-8859-1; format=flowed > > Can you send in as much information on that system as you can? Add in > cards, bios version, even who made the case if possible. Thanks. > > Arthur > > I'm away from my office now, and I cannot find these details now, but I include below the output from lspci in case that can be useful. Next week I will try to get this information and post it back here. > ------------------------------ > > Message: 2 > Date: Fri, 10 Jul 2009 11:15:03 -0700 > From: Greg Kurtzer > Subject: Re: [Warewulf] Initial problems with Caos NSA - Perceus > To: The Warewulf Cluster Toolkit > Message-ID: > <571f1a060907101115s693c33ebwb689f20f0ceb3c11 at mail.gmail.com> > Content-Type: text/plain; charset=ISO-8859-1 > > And in addition to what Arthur mentioned, the output of: > > # lspci; lspci -n > > on a system with the same network interface would be very helpful in > resolving this. > > Thank you. > > Greg > > Here it goes both outputs in the same system where I was trying to install Perceus (momentarily I had to install Rocks in it... but I will come back to Perceus if I can). [root at pvfs2-compute-0-0 ~]# lspci 00:00.0 Memory controller: nVidia Corporation CK804 Memory Controller (rev a3) 00:01.0 ISA bridge: nVidia Corporation CK804 ISA Bridge (rev a3) 00:01.1 SMBus: nVidia Corporation CK804 SMBus (rev a2) 00:02.0 USB Controller: nVidia Corporation CK804 USB Controller (rev a2) 00:02.1 USB Controller: nVidia Corporation CK804 USB Controller (rev a3) 00:06.0 IDE interface: nVidia Corporation CK804 IDE (rev f2) 00:07.0 IDE interface: nVidia Corporation CK804 Serial ATA Controller (rev f3) 00:08.0 IDE interface: nVidia Corporation CK804 Serial ATA Controller (rev f3) 00:09.0 PCI bridge: nVidia Corporation CK804 PCI Bridge (rev a2) 00:0e.0 PCI bridge: nVidia Corporation CK804 PCIE Bridge (rev a3) 00:18.0 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] HyperTransport Technology Configuration 00:18.1 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] Address Map 00:18.2 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] DRAM Controller 00:18.3 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] Miscellaneous Control 00:19.0 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] HyperTransport Technology Configuration 00:19.1 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] Address Map 00:19.2 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] DRAM Controller 00:19.3 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] Miscellaneous Control 01:07.0 VGA compatible controller: ATI Technologies Inc Rage XL (rev 27) 08:0a.0 PCI bridge: Advanced Micro Devices [AMD] AMD-8131 PCI-X Bridge (rev 12) 08:0a.1 PIC: Advanced Micro Devices [AMD] AMD-8131 PCI-X IOAPIC (rev 01) 08:0b.0 PCI bridge: Advanced Micro Devices [AMD] AMD-8131 PCI-X Bridge (rev 12) 08:0b.1 PIC: Advanced Micro Devices [AMD] AMD-8131 PCI-X IOAPIC (rev 01) 0a:09.0 Ethernet controller: Broadcom Corporation NetXtreme BCM5704 Gigabit Ethernet (rev 03) 0a:09.1 Ethernet controller: Broadcom Corporation NetXtreme BCM5704 Gigabit Ethernet (rev 03) [root at pvfs2-compute-0-0 ~]# [root at pvfs2-compute-0-0 ~]# lspci -n 00:00.0 0580: 10de:005e (rev a3) 00:01.0 0601: 10de:0051 (rev a3) 00:01.1 0c05: 10de:0052 (rev a2) 00:02.0 0c03: 10de:005a (rev a2) 00:02.1 0c03: 10de:005b (rev a3) 00:06.0 0101: 10de:0053 (rev f2) 00:07.0 0101: 10de:0054 (rev f3) 00:08.0 0101: 10de:0055 (rev f3) 00:09.0 0604: 10de:005c (rev a2) 00:0e.0 0604: 10de:005d (rev a3) 00:18.0 0600: 1022:1100 00:18.1 0600: 1022:1101 00:18.2 0600: 1022:1102 00:18.3 0600: 1022:1103 00:19.0 0600: 1022:1100 00:19.1 0600: 1022:1101 00:19.2 0600: 1022:1102 00:19.3 0600: 1022:1103 01:07.0 0300: 1002:4752 (rev 27) 08:0a.0 0604: 1022:7450 (rev 12) 08:0a.1 0800: 1022:7451 (rev 01) 08:0b.0 0604: 1022:7450 (rev 12) 08:0b.1 0800: 1022:7451 (rev 01) 0a:09.0 0200: 14e4:1648 (rev 03) 0a:09.1 0200: 14e4:1648 (rev 03) [root at pvfs2-compute-0-0 ~]# Thanks a lot, ?ngel de Vicente -- +---------------------------------------------+ | | | http://www.iac.es/galeria/angelv/ | | | | High Performance Computing Support PostDoc | | Instituto de Astrof?sica de Canarias | | | +---------------------------------------------+ --------------------------------------------------------------------------------------------- ADVERTENCIA: Sobre la privacidad y cumplimiento de la Ley de Protecci?n de Datos, acceda a http://www.iac.es/disclaimer.php WARNING: For more information on privacy and fulfilment of the Law concerning the Protection of Data, consult http://www.iac.es/disclaimer.php?lang=en From gmkurtzer at gmail.com Fri Jul 17 07:20:39 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Fri, 17 Jul 2009 07:20:39 -0700 Subject: [Warewulf] Initial problems with Caos NSA - Perceus In-Reply-To: <4A603242.1010706@iac.es> References: <4A603242.1010706@iac.es> Message-ID: <571f1a060907170720r1aa85871r9624cfea5b50ed5b@mail.gmail.com> Hello, Those Broadcoms require the tg3 module which is supported just fine with Perceus so the problem isn't what I was thinking. What motherboards/systems are these? Greg On Fri, Jul 17, 2009 at 1:11 AM, ?ngel de Vicente wrote: > Hi, > > sorry for the late reply to this... It was a VERY busy week... >> ---------------------------------------------------------------------- >> >> Message: 1 >> Date: Fri, 10 Jul 2009 06:48:13 -0700 >> From: "Arthur A. Stevens" >> Subject: Re: [Warewulf] Initial problems with Caos NSA - Perceus >> To: The Warewulf Cluster Toolkit >> Message-ID: <4A57469D.6060001 at Infiscale.com> >> Content-Type: text/plain; charset=ISO-8859-1; format=flowed >> >> Can you send in as much information on that system as you can? Add in >> cards, bios version, even who made the case if possible. Thanks. >> >> Arthur >> >> > I'm away from my office now, and I cannot find these details now, but I > include below the output from lspci in case that can be useful. Next > week I will try to get this information and post it back here. > >> ------------------------------ >> >> Message: 2 >> Date: Fri, 10 Jul 2009 11:15:03 -0700 >> From: Greg Kurtzer >> Subject: Re: [Warewulf] Initial problems with Caos NSA - Perceus >> To: The Warewulf Cluster Toolkit >> Message-ID: >> ? ? ? <571f1a060907101115s693c33ebwb689f20f0ceb3c11 at mail.gmail.com> >> Content-Type: text/plain; charset=ISO-8859-1 >> >> And in addition to what Arthur mentioned, the output of: >> >> # lspci; lspci -n >> >> on a system with the same network interface would be very helpful in >> resolving this. >> >> Thank you. >> >> Greg >> >> > Here it goes both outputs in the same system where I was trying to > install Perceus (momentarily I had to install Rocks in it... but I will > come back to Perceus if I can). > > [root at pvfs2-compute-0-0 ~]# lspci > 00:00.0 Memory controller: nVidia Corporation CK804 Memory Controller > (rev a3) > 00:01.0 ISA bridge: nVidia Corporation CK804 ISA Bridge (rev a3) > 00:01.1 SMBus: nVidia Corporation CK804 SMBus (rev a2) > 00:02.0 USB Controller: nVidia Corporation CK804 USB Controller (rev a2) > 00:02.1 USB Controller: nVidia Corporation CK804 USB Controller (rev a3) > 00:06.0 IDE interface: nVidia Corporation CK804 IDE (rev f2) > 00:07.0 IDE interface: nVidia Corporation CK804 Serial ATA Controller > (rev f3) > 00:08.0 IDE interface: nVidia Corporation CK804 Serial ATA Controller > (rev f3) > 00:09.0 PCI bridge: nVidia Corporation CK804 PCI Bridge (rev a2) > 00:0e.0 PCI bridge: nVidia Corporation CK804 PCIE Bridge (rev a3) > 00:18.0 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] > HyperTransport Technology Configuration > 00:18.1 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] > Address Map > 00:18.2 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] > DRAM Controller > 00:18.3 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] > Miscellaneous Control > 00:19.0 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] > HyperTransport Technology Configuration > 00:19.1 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] > Address Map > 00:19.2 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] > DRAM Controller > 00:19.3 Host bridge: Advanced Micro Devices [AMD] K8 [Athlon64/Opteron] > Miscellaneous Control > 01:07.0 VGA compatible controller: ATI Technologies Inc Rage XL (rev 27) > 08:0a.0 PCI bridge: Advanced Micro Devices [AMD] AMD-8131 PCI-X Bridge > (rev 12) > 08:0a.1 PIC: Advanced Micro Devices [AMD] AMD-8131 PCI-X IOAPIC (rev 01) > 08:0b.0 PCI bridge: Advanced Micro Devices [AMD] AMD-8131 PCI-X Bridge > (rev 12) > 08:0b.1 PIC: Advanced Micro Devices [AMD] AMD-8131 PCI-X IOAPIC (rev 01) > 0a:09.0 Ethernet controller: Broadcom Corporation NetXtreme BCM5704 > Gigabit Ethernet (rev 03) > 0a:09.1 Ethernet controller: Broadcom Corporation NetXtreme BCM5704 > Gigabit Ethernet (rev 03) > [root at pvfs2-compute-0-0 ~]# > > > [root at pvfs2-compute-0-0 ~]# lspci -n > 00:00.0 0580: 10de:005e (rev a3) > 00:01.0 0601: 10de:0051 (rev a3) > 00:01.1 0c05: 10de:0052 (rev a2) > 00:02.0 0c03: 10de:005a (rev a2) > 00:02.1 0c03: 10de:005b (rev a3) > 00:06.0 0101: 10de:0053 (rev f2) > 00:07.0 0101: 10de:0054 (rev f3) > 00:08.0 0101: 10de:0055 (rev f3) > 00:09.0 0604: 10de:005c (rev a2) > 00:0e.0 0604: 10de:005d (rev a3) > 00:18.0 0600: 1022:1100 > 00:18.1 0600: 1022:1101 > 00:18.2 0600: 1022:1102 > 00:18.3 0600: 1022:1103 > 00:19.0 0600: 1022:1100 > 00:19.1 0600: 1022:1101 > 00:19.2 0600: 1022:1102 > 00:19.3 0600: 1022:1103 > 01:07.0 0300: 1002:4752 (rev 27) > 08:0a.0 0604: 1022:7450 (rev 12) > 08:0a.1 0800: 1022:7451 (rev 01) > 08:0b.0 0604: 1022:7450 (rev 12) > 08:0b.1 0800: 1022:7451 (rev 01) > 0a:09.0 0200: 14e4:1648 (rev 03) > 0a:09.1 0200: 14e4:1648 (rev 03) > [root at pvfs2-compute-0-0 ~]# > > Thanks a lot, > ?ngel de Vicente > > -- > +---------------------------------------------+ > | ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? | > | http://www.iac.es/galeria/angelv/ ? ? ? ? ? | > | ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? | > | High Performance Computing Support PostDoc ?| > | Instituto de Astrof?sica de Canarias ? ? ? ?| > | ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? | > +---------------------------------------------+ > > --------------------------------------------------------------------------------------------- > ADVERTENCIA: Sobre la privacidad y cumplimiento de la Ley de Protecci?n de Datos, acceda a http://www.iac.es/disclaimer.php > WARNING: For more information on privacy and fulfilment of the Law concerning the Protection of Data, consult http://www.iac.es/disclaimer.php?lang=en > > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From gmkurtzer at gmail.com Fri Jul 17 07:22:55 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Fri, 17 Jul 2009 07:22:55 -0700 Subject: [Warewulf] links on vnfs capsules In-Reply-To: References: Message-ID: <571f1a060907170722x4a8d6376r6953b78d63dd5db3@mail.gmail.com> Hello, Yes as you noted the mount command is very fast indeed and the hybridization is done when the capsule is unmounted. Check the file /etc/perceus/vnfs/[vnfs name]/hybridize. I bet those directories are listed there. Good luck! Greg On Thu, Jul 16, 2009 at 7:17 PM, Gary Leong wrote: > Actually, I see when I do a > > perceus vnfs mount capsule > > It happens really quick, to quick.? It's not uncompressing my capsule.? It's > linking to my a copy of chroot.? Therefore, that's why the links aren't > showing, ah?? I have to dismantle the capsule myself to see the links, ah? > > Gary > > On Thu, Jul 16, 2009 at 7:15 PM, Gary Leong wrote: >> >> I'm not using the hybridization model, but I'm getting these links when I >> boot up.? Hmm...where do the links come from?? When I do a vnfs mount, I >> don't see the links, but libraries I need.? But when I boot up, something >> blows away my /usr/share /usr/src and /usr/include and replaces it with this >> links that are blinking because they don't connect to anywhere.?? A lot of >> things break because of this. >> >> Any help would be appreciated.? Thanks. >> >> gary >> >> lrwxrwxrwx? 1 root root 59 Jun 28 06:00 include -> >> /var/lib/perceus//vnfs/Centos-5.3.x86_64/rootfs/usr/include >> drwxr-xr-x? 4 root root? 0 Jun 28 06:00 kerberos >> drwxr-xr-x 13 root root? 0 Jun 28 06:00 lib >> drwxr-xr-x 19 root root? 0 Jun 28 06:00 lib64 >> drwxr-xr-x? 5 root root? 0 Jun 28 06:00 libexec >> drwxr-xr-x 12 root root? 0 Jun 28 06:00 local >> drwxr-xr-x? 3 root root? 0 Jun 28 06:00 man >> drwxr-xr-x? 2 root root? 0 Jun 28 06:00 sbin >> lrwxrwxrwx? 1 root root 57 Jun 28 06:00 share -> >> /var/lib/perceus//vnfs/Centos-5.3.x86_64/rootfs/usr/share >> lrwxrwxrwx? 1 root root 55 Jun 28 06:00 src -> >> /var/lib/perceus//vnfs/Centos-5.3.x86_64/rootfs/usr/src >> lrwxrwxrwx? 1 root root 10 Jun 28 06:00 tmp -> ../var/tmp >> > > > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf > > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From gmkurtzer at gmail.com Fri Jul 17 07:24:43 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Fri, 17 Jul 2009 07:24:43 -0700 Subject: [Warewulf] Turning off hybridization in Perceus In-Reply-To: References: Message-ID: <571f1a060907170724q4e8168a3m74e01cbfcb312a67@mail.gmail.com> Hello, I like that idea. Probably should be a feature update for Perceus 2.0 (in the works now). Thanks, Greg On Mon, Jul 13, 2009 at 12:05 PM, Bernard Li wrote: > Hi guys: > > Currently, when configuring a VNFS image (either by > chroot2stateless.sh or by perceus vnfs configure), it would ask > whether you want to enable hybridization or not. ?If you answer no, it > would simply not set up fstab entry for the statedir. > > The issue is that the hybridize file is still created in your VNFS > image dir and the default directories are still "hybridized", so the > end result is that a node when booted against that image, would have > dangling symlinks to these directories. > > The current workaround is simply to uncomment the directories in the > hybridize file, but it would be nice if there was a way to simply turn > hybridization off (maybe through a config option?) > > Thanks, > > Bernard > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From bernard at vanhpc.org Fri Jul 17 10:55:18 2009 From: bernard at vanhpc.org (Bernard Li) Date: Fri, 17 Jul 2009 10:55:18 -0700 Subject: [Warewulf] links on vnfs capsules In-Reply-To: References: Message-ID: Hi Gary: See my other post regarding hybridization: http://lists.caosity.org/pipermail/warewulf/2009-July/003825.html Basically even though you are "not using" hybridization, if you still have uncommented lines in the hybridize file in your image, Perceus would still hybridize it. The workaround is to simply comment the lines in the file. In general, enabling hybridization for a stateless node is a good thing since it saves memory usage -- is there a particular reason why you do not want this enabled? Cheers, Bernard On Thu, Jul 16, 2009 at 7:17 PM, Gary Leong wrote: > Actually, I see when I do a > > perceus vnfs mount capsule > > It happens really quick, to quick.? It's not uncompressing my capsule.? It's > linking to my a copy of chroot.? Therefore, that's why the links aren't > showing, ah?? I have to dismantle the capsule myself to see the links, ah? > > Gary > > On Thu, Jul 16, 2009 at 7:15 PM, Gary Leong wrote: >> >> I'm not using the hybridization model, but I'm getting these links when I >> boot up.? Hmm...where do the links come from?? When I do a vnfs mount, I >> don't see the links, but libraries I need.? But when I boot up, something >> blows away my /usr/share /usr/src and /usr/include and replaces it with this >> links that are blinking because they don't connect to anywhere.?? A lot of >> things break because of this. >> >> Any help would be appreciated.? Thanks. >> >> gary >> >> lrwxrwxrwx? 1 root root 59 Jun 28 06:00 include -> >> /var/lib/perceus//vnfs/Centos-5.3.x86_64/rootfs/usr/include >> drwxr-xr-x? 4 root root? 0 Jun 28 06:00 kerberos >> drwxr-xr-x 13 root root? 0 Jun 28 06:00 lib >> drwxr-xr-x 19 root root? 0 Jun 28 06:00 lib64 >> drwxr-xr-x? 5 root root? 0 Jun 28 06:00 libexec >> drwxr-xr-x 12 root root? 0 Jun 28 06:00 local >> drwxr-xr-x? 3 root root? 0 Jun 28 06:00 man >> drwxr-xr-x? 2 root root? 0 Jun 28 06:00 sbin >> lrwxrwxrwx? 1 root root 57 Jun 28 06:00 share -> >> /var/lib/perceus//vnfs/Centos-5.3.x86_64/rootfs/usr/share >> lrwxrwxrwx? 1 root root 55 Jun 28 06:00 src -> >> /var/lib/perceus//vnfs/Centos-5.3.x86_64/rootfs/usr/src >> lrwxrwxrwx? 1 root root 10 Jun 28 06:00 tmp -> ../var/tmp >> > > > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf > > From gwleong at gmail.com Fri Jul 17 11:42:52 2009 From: gwleong at gmail.com (Gary Leong) Date: Fri, 17 Jul 2009 11:42:52 -0700 Subject: [Warewulf] links on vnfs capsules In-Reply-To: References: Message-ID: Since going back into the private sector, my having an NFS mount provides one single point of failure. For parallel databases/clusters, we can have one node go down, but if all of them go down because the nfs server goes down,it's really bad news. It's not worth saving memory because the risk is too high. I'm using the vnfs capsules first to provision onto to bare-metal, and then we will consider something like diskless. In general, my diskless solution would be to hybridize toward a local disk that's rsync with / in memory. Anyhow, I did comment them out and it works. Thanks a bunch! Gary On Fri, Jul 17, 2009 at 10:55 AM, Bernard Li wrote: > Hi Gary: > > See my other post regarding hybridization: > > http://lists.caosity.org/pipermail/warewulf/2009-July/003825.html > > Basically even though you are "not using" hybridization, if you still > have uncommented lines in the hybridize file in your image, Perceus > would still hybridize it. The workaround is to simply comment the > lines in the file. > > In general, enabling hybridization for a stateless node is a good > thing since it saves memory usage -- is there a particular reason why > you do not want this enabled? > > Cheers, > > Bernard > > On Thu, Jul 16, 2009 at 7:17 PM, Gary Leong wrote: > > Actually, I see when I do a > > > > perceus vnfs mount capsule > > > > It happens really quick, to quick. It's not uncompressing my capsule. > It's > > linking to my a copy of chroot. Therefore, that's why the links aren't > > showing, ah? I have to dismantle the capsule myself to see the links, > ah? > > > > Gary > > > > On Thu, Jul 16, 2009 at 7:15 PM, Gary Leong wrote: > >> > >> I'm not using the hybridization model, but I'm getting these links when > I > >> boot up. Hmm...where do the links come from? When I do a vnfs mount, I > >> don't see the links, but libraries I need. But when I boot up, > something > >> blows away my /usr/share /usr/src and /usr/include and replaces it with > this > >> links that are blinking because they don't connect to anywhere. A lot > of > >> things break because of this. > >> > >> Any help would be appreciated. Thanks. > >> > >> gary > >> > >> lrwxrwxrwx 1 root root 59 Jun 28 06:00 include -> > >> /var/lib/perceus//vnfs/Centos-5.3.x86_64/rootfs/usr/include > >> drwxr-xr-x 4 root root 0 Jun 28 06:00 kerberos > >> drwxr-xr-x 13 root root 0 Jun 28 06:00 lib > >> drwxr-xr-x 19 root root 0 Jun 28 06:00 lib64 > >> drwxr-xr-x 5 root root 0 Jun 28 06:00 libexec > >> drwxr-xr-x 12 root root 0 Jun 28 06:00 local > >> drwxr-xr-x 3 root root 0 Jun 28 06:00 man > >> drwxr-xr-x 2 root root 0 Jun 28 06:00 sbin > >> lrwxrwxrwx 1 root root 57 Jun 28 06:00 share -> > >> /var/lib/perceus//vnfs/Centos-5.3.x86_64/rootfs/usr/share > >> lrwxrwxrwx 1 root root 55 Jun 28 06:00 src -> > >> /var/lib/perceus//vnfs/Centos-5.3.x86_64/rootfs/usr/src > >> lrwxrwxrwx 1 root root 10 Jun 28 06:00 tmp -> ../var/tmp > >> > > > > > > _______________________________________________ > > Warewulf mailing list > > Warewulf at caoslinux.org > > http://lists.caosity.org/mailman/listinfo/warewulf > > > > > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf > -------------- next part -------------- An HTML attachment was scrubbed... URL: http://altruistic.infiscale.org/pipermail/perceus/attachments/20090717/0f9cf010/attachment.html From gmkurtzer at gmail.com Fri Jul 17 14:44:46 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Fri, 17 Jul 2009 14:44:46 -0700 Subject: [Warewulf] links on vnfs capsules In-Reply-To: References: Message-ID: <571f1a060907171444g1420591eydc46ce1477c8dc28@mail.gmail.com> How are you sharing the user's home directories? If that goes away the cluster is as good as done (at least for a traditional HPC type cluster). What we suggest is put the user's home directories and the Perceus statedir on the same device, to consolidate and minimize points of failures. Then get a really solid enterprise class NFS solution for both. Good luck! Greg On Fri, Jul 17, 2009 at 11:42 AM, Gary Leong wrote: > Since going back into the private sector, my having an NFS mount provides > one single point of failure. For parallel databases/clusters, we can have > one node go down, but if all of them go down because the nfs server goes > down,it's really bad news.? It's not worth saving memory because the risk is > too high. I'm using the vnfs capsules first to provision onto to bare-metal, > and then we will consider something like diskless.? In general, my diskless > solution would be? to hybridize toward a local disk that's rsync with / in > memory. > > Anyhow, I did comment them out and it works.?? Thanks a bunch! > > Gary > > On Fri, Jul 17, 2009 at 10:55 AM, Bernard Li wrote: >> >> Hi Gary: >> >> See my other post regarding hybridization: >> >> http://lists.caosity.org/pipermail/warewulf/2009-July/003825.html >> >> Basically even though you are "not using" hybridization, if you still >> have uncommented lines in the hybridize file in your image, Perceus >> would still hybridize it. ?The workaround is to simply comment the >> lines in the file. >> >> In general, enabling hybridization for a stateless node is a good >> thing since it saves memory usage -- is there a particular reason why >> you do not want this enabled? >> >> Cheers, >> >> Bernard >> >> On Thu, Jul 16, 2009 at 7:17 PM, Gary Leong wrote: >> > Actually, I see when I do a >> > >> > perceus vnfs mount capsule >> > >> > It happens really quick, to quick.? It's not uncompressing my capsule. >> > It's >> > linking to my a copy of chroot.? Therefore, that's why the links aren't >> > showing, ah?? I have to dismantle the capsule myself to see the links, >> > ah? >> > >> > Gary >> > >> > On Thu, Jul 16, 2009 at 7:15 PM, Gary Leong wrote: >> >> >> >> I'm not using the hybridization model, but I'm getting these links when >> >> I >> >> boot up.? Hmm...where do the links come from?? When I do a vnfs mount, >> >> I >> >> don't see the links, but libraries I need.? But when I boot up, >> >> something >> >> blows away my /usr/share /usr/src and /usr/include and replaces it with >> >> this >> >> links that are blinking because they don't connect to anywhere.?? A lot >> >> of >> >> things break because of this. >> >> >> >> Any help would be appreciated.? Thanks. >> >> >> >> gary >> >> >> >> lrwxrwxrwx? 1 root root 59 Jun 28 06:00 include -> >> >> /var/lib/perceus//vnfs/Centos-5.3.x86_64/rootfs/usr/include >> >> drwxr-xr-x? 4 root root? 0 Jun 28 06:00 kerberos >> >> drwxr-xr-x 13 root root? 0 Jun 28 06:00 lib >> >> drwxr-xr-x 19 root root? 0 Jun 28 06:00 lib64 >> >> drwxr-xr-x? 5 root root? 0 Jun 28 06:00 libexec >> >> drwxr-xr-x 12 root root? 0 Jun 28 06:00 local >> >> drwxr-xr-x? 3 root root? 0 Jun 28 06:00 man >> >> drwxr-xr-x? 2 root root? 0 Jun 28 06:00 sbin >> >> lrwxrwxrwx? 1 root root 57 Jun 28 06:00 share -> >> >> /var/lib/perceus//vnfs/Centos-5.3.x86_64/rootfs/usr/share >> >> lrwxrwxrwx? 1 root root 55 Jun 28 06:00 src -> >> >> /var/lib/perceus//vnfs/Centos-5.3.x86_64/rootfs/usr/src >> >> lrwxrwxrwx? 1 root root 10 Jun 28 06:00 tmp -> ../var/tmp >> >> >> > >> > >> > _______________________________________________ >> > Warewulf mailing list >> > Warewulf at caoslinux.org >> > http://lists.caosity.org/mailman/listinfo/warewulf >> > >> > >> _______________________________________________ >> Warewulf mailing list >> Warewulf at caoslinux.org >> http://lists.caosity.org/mailman/listinfo/warewulf > > > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf > > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From bernard at vanhpc.org Mon Jul 20 17:43:45 2009 From: bernard at vanhpc.org (Bernard Li) Date: Mon, 20 Jul 2009 17:43:45 -0700 Subject: [Warewulf] Bug with stateless provisioning of nash-based initrd with Perceus 1.6.0 Message-ID: Hi: I created a VNFS image using chroot2stateless.sh from Perceus 1.6.0 -- the resulting 89-vnfsinit.sh script checks to see if /sbin/nash exists and if it does execute some nash-specific code. It looks like the nash-specific code isn't working, as the node kernel panics when it boots: Red Hat nash version 5.1.19.6 Failed to read startup file /sysroot Here's the code in question: --- echo "if [ -x /sbin/nash ]; then" >> \$DESTDIR/init echo "mkdir /sysroot" >> \$DESTDIR/init echo "rsync -aH --exclude '/proc/*' --exclude '/sys/*' --exclude '/sysroot' /* /sysroot/" >> \$DESTDIR/init echo "echo switchroot | /sbin/nash --force /sysroot" >> \$DESTDIR/init --- If the code snippet is commented out, the node boots up fine. I tested this with CentOS 5.3 i386. Thanks, Bernard From gmkurtzer at gmail.com Mon Jul 20 19:42:11 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Mon, 20 Jul 2009 19:42:11 -0700 Subject: [Warewulf] Bug with stateless provisioning of nash-based initrd with Perceus 1.6.0 In-Reply-To: References: Message-ID: <571f1a060907201942y303facb4y679a4673da88ced3@mail.gmail.com> Hiya Bernard, That code wasnt tested yet, but the goal of it was to facilitate the switchroot onto the new tmpfs (instead of ramfs as you pointed out earlier). My guess is that it is still running on the ramfs which is what we were hoping to avoid. We will check into this shortly, but if you find a solution before us feel free to let us know! Thanks! Greg On Mon, Jul 20, 2009 at 5:43 PM, Bernard Li wrote: > Hi: > > I created a VNFS image using chroot2stateless.sh from Perceus 1.6.0 -- > the resulting 89-vnfsinit.sh script checks to see if /sbin/nash exists > and if it does execute some nash-specific code. > > It looks like the nash-specific code isn't working, as the node kernel > panics when it boots: > > Red Hat nash version 5.1.19.6 > Failed to read startup file /sysroot > > Here's the code in question: > > --- > echo "if [ -x /sbin/nash ]; then" >> \$DESTDIR/init > echo "mkdir /sysroot" >> \$DESTDIR/init > echo "rsync -aH --exclude '/proc/*' --exclude '/sys/*' --exclude > '/sysroot' /* /sysroot/" >> \$DESTDIR/init > echo "echo switchroot | /sbin/nash --force /sysroot" >> \$DESTDIR/init > --- > > If the code snippet is commented out, the node boots up fine. > > I tested this with CentOS 5.3 i386. > > Thanks, > > Bernard > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From jbh at broadinstitute.org Tue Jul 21 12:40:39 2009 From: jbh at broadinstitute.org (John Hanks) Date: Tue, 21 Jul 2009 15:40:39 -0400 Subject: [Warewulf] using ipaddr module to bring a nic up with no config. Message-ID: I'm trying to bring up a nodes nic with no configuration so it can then have one or more tagged VLANs configured on it. I've tried numerous permutations but have been unsuccessful. Does anyone have a working configuration for doing this with the ipaddr module? Basically I'd like this to work as I expect it too node* eth0(PEERDNS=no):[default]/[default] eth1(MTU=9000):none and just bring eth1 up with no assigned address and with and mtu of 9000. Instead I get this ifcfg-eth1 file which includes aspects of the eth0 config: DEVICE=eth1 ONBOOT=yes TYPE=ethernet PROTO=static BOOTPROTO=static IPADDR=10.128.8.185 NETMASK=255.255.0.0 GATEWAY= DEVICE=eth1 ONBOOT=yes TYPE=ethernet PROTO=static BOOTPROTO=static IPADDR=none NETMASK= GATEWAY= MTU=9000 Thanks, jbh From gmkurtzer at gmail.com Tue Jul 21 16:09:56 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Tue, 21 Jul 2009 16:09:56 -0700 Subject: [Warewulf] using ipaddr module to bring a nic up with no config. In-Reply-To: References: Message-ID: <571f1a060907211609i662452ebnb21a32de26b96782@mail.gmail.com> Hi John! What version of Perceus is this? Have you tried removing the none? With this config: test* eth0(PEERDNS=no):[default]/[default] eth1(MTU=9000): I get: DEVICE=eth1 ONBOOT=yes TYPE=ethernet MTU=9000 There was some fixes that you may need that went into a SCM update: # perceus --version Version=1.5.2 Build=2072 Thanks, Greg On Tue, Jul 21, 2009 at 12:40 PM, John Hanks wrote: > I'm trying to bring up a nodes nic with no configuration so it can > then have one or more tagged VLANs configured on it. I've tried > numerous permutations but have been unsuccessful. Does anyone have a > working configuration for doing this with the ipaddr module? Basically > I'd like this to work as I expect it too > > node* eth0(PEERDNS=no):[default]/[default] eth1(MTU=9000):none > > and just bring eth1 up with no assigned address and with and mtu of > 9000. Instead I get this ifcfg-eth1 file which includes aspects of the > eth0 config: > > DEVICE=eth1 > ONBOOT=yes > TYPE=ethernet > PROTO=static > BOOTPROTO=static > IPADDR=10.128.8.185 > NETMASK=255.255.0.0 > GATEWAY= > DEVICE=eth1 > ONBOOT=yes > TYPE=ethernet > PROTO=static > BOOTPROTO=static > IPADDR=none > NETMASK= > GATEWAY= > MTU=9000 > > > Thanks, > > jbh > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From jbh at broadinstitute.org Tue Jul 21 17:31:29 2009 From: jbh at broadinstitute.org (John Hanks) Date: Tue, 21 Jul 2009 20:31:29 -0400 Subject: [Warewulf] using ipaddr module to bring a nic up with no config. In-Reply-To: <571f1a060907211609i662452ebnb21a32de26b96782@mail.gmail.com> References: <571f1a060907211609i662452ebnb21a32de26b96782@mail.gmail.com> Message-ID: Sorry, I left out much pertinent information in my haste to click send before rushing out of the office. [root at perceus ~]# perceus --version Version=1.5.2 Build=0.2092 with a line like yours I still get an odd ifcfg-eth1 file: DEVICE=eth1 ONBOOT=yes TYPE=ethernet PROTO=static BOOTPROTO=static IPADDR=10.128.8.184 NETMASK=255.255.0.0 GATEWAY= DEVICE=eth1 ONBOOT=yes TYPE=ethernet MTU=9000 eth0 has a consistently odd pattern: DEVICE=eth0 ONBOOT=yes TYPE=ethernet PROTO=static BOOTPROTO=static IPADDR=10.128.8.184 NETMASK=255.255.0.0 GATEWAY= DEVICE=eth0 ONBOOT=yes TYPE=ethernet PROTO=static BOOTPROTO=static IPADDR=10.128.8.184 NETMASK=255.255.0.0 GATEWAY= PEERDNS=no as if both devices get configured twice. Thanks, jbh On Tue, Jul 21, 2009 at 7:09 PM, Greg Kurtzer wrote: > Hi John! > > What version of Perceus is this? > > Have you tried removing the none? > > With this config: > > test* eth0(PEERDNS=no):[default]/[default] eth1(MTU=9000): > > I get: > > DEVICE=eth1 > ONBOOT=yes > TYPE=ethernet > MTU=9000 > > There was some fixes that you may need that went into a SCM update: > > # perceus --version > Version=1.5.2 > Build=2072 > > > Thanks, > Greg > > > On Tue, Jul 21, 2009 at 12:40 PM, John Hanks wrote: >> I'm trying to bring up a nodes nic with no configuration so it can >> then have one or more tagged VLANs configured on it. I've tried >> numerous permutations but have been unsuccessful. Does anyone have a >> working configuration for doing this with the ipaddr module? Basically >> I'd like this to work as I expect it too >> >> node* eth0(PEERDNS=no):[default]/[default] eth1(MTU=9000):none >> >> and just bring eth1 up with no assigned address and with and mtu of >> 9000. Instead I get this ifcfg-eth1 file which includes aspects of the >> eth0 config: >> >> DEVICE=eth1 >> ONBOOT=yes >> TYPE=ethernet >> PROTO=static >> BOOTPROTO=static >> IPADDR=10.128.8.185 >> NETMASK=255.255.0.0 >> GATEWAY= >> DEVICE=eth1 >> ONBOOT=yes >> TYPE=ethernet >> PROTO=static >> BOOTPROTO=static >> IPADDR=none >> NETMASK= >> GATEWAY= >> MTU=9000 >> >> >> Thanks, >> >> jbh >> _______________________________________________ >> Warewulf mailing list >> Warewulf at caoslinux.org >> http://lists.caosity.org/mailman/listinfo/warewulf >> > > > > -- > Greg Kurtzer > http://www.infiscale.com/ > http://www.perceus.org/ > http://www.caoslinux.org/ > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf > From gwleong at gmail.com Wed Jul 22 14:44:51 2009 From: gwleong at gmail.com (Gary Leong) Date: Wed, 22 Jul 2009 14:44:51 -0700 Subject: [Warewulf] perceus initiation Message-ID: Is there a manual way to initialize Perceus? I want to script out the installation of Perceus rather than have it prompt me for questions. The first time you run perceus, it complains that perceus needs to be initalize even though I setup the MySQL database, and configured perceus.conf, defaults.conf, and dnsmasq.conf. If I run the initialize scripts, it clobbers these files, and I have to do them all over. There are are some bugs with initialization scripts. Should I list them here or is there a bug tracker you guys used? Thanks! Gary -------------- next part -------------- An HTML attachment was scrubbed... URL: http://altruistic.infiscale.org/pipermail/perceus/attachments/20090722/1f5aaf4b/attachment.html From bernard at vanhpc.org Wed Jul 22 17:07:00 2009 From: bernard at vanhpc.org (Bernard Li) Date: Wed, 22 Jul 2009 17:07:00 -0700 Subject: [Warewulf] Bug with stateless provisioning of nash-based initrd with Perceus 1.6.0 In-Reply-To: <571f1a060907201942y303facb4y679a4673da88ced3@mail.gmail.com> References: <571f1a060907201942y303facb4y679a4673da88ced3@mail.gmail.com> Message-ID: Hi Greg: On Mon, Jul 20, 2009 at 7:42 PM, Greg Kurtzer wrote: > That code wasnt tested yet, but the goal of it was to facilitate the > switchroot onto the new tmpfs (instead of ramfs as you pointed out > earlier). My guess is that it is still running on the ramfs which is > what we were hoping to avoid. I have found a solution, please see the attached patch. When switchroot runs, it automatically umounts old /proc, /sys and /dev, so I set those up to prevent error messages. Also, /sysroot needs to be mounted separately from / (in this case I mounted it as tmpfs). switchroot cannot be called by a subprocess, therefore the only way to get it working is to call nash via exec. You would think that "echo switchroot | exec /sbin/nash" would work, but it doesn't, so the only way is to pass it a file with the switchroot command and that's what I've done, and it works now. Special thanks to pjones in #anaconda for his help on figuring this out. Perhaps we should implement a small program to facilitate switchroot instead of relying on nash... because right now this would work for nash based systems (i.e. Fedora, Red Hat, CentOS, etc.) Oh and just to confirm that the new code puts the OS in tmpfs instead of ramfs: [/sbin/nash does not exist] # df -h / Filesystem Size Used Avail Use% Mounted on none 0 0 0 - / [/sbin/nash exists] # df -h / Filesystem Size Used Avail Use% Mounted on none 1014M 152M 862M 15% / Cheers, Bernard -------------- next part -------------- A non-text attachment was scrubbed... Name: chroot2stateless.sh.patch Type: text/x-patch Size: 1277 bytes Desc: not available Url : http://altruistic.infiscale.org/pipermail/perceus/attachments/20090722/bf6a0083/attachment.bin From bernard at vanhpc.org Wed Jul 22 17:40:24 2009 From: bernard at vanhpc.org (Bernard Li) Date: Wed, 22 Jul 2009 17:40:24 -0700 Subject: [Warewulf] perceus initiation In-Reply-To: References: Message-ID: Hi Gary: On Wed, Jul 22, 2009 at 2:44 PM, Gary Leong wrote: > Is there a manual way to initialize Perceus?? I want to script out the > installation of Perceus rather than have it prompt me for questions.? The > first time you run perceus, it complains that perceus needs to be initalize > even though I setup the MySQL database, and configured perceus.conf, > defaults.conf, and dnsmasq.conf.? If I run the initialize scripts, it > clobbers these files, and I have to do them all over. If memory serves, as long as /var/lib/perceus/database exists, then Perceus would treat the system as initialized. It sounds like you have already manually "initialized" Perceus -- does it not work after your manual configuration? For me, I would prime the Perceus setup manually and then call these commands as necessary: perceus configure nfs perceus configure sshkeys Another option is to call perceus init by using an expect script. But I think what would be even better is for Perceus to support a way of doing automated installation -- I would give my +1 behind that. > There are are some bugs with initialization scripts.? Should I list them > here or is there a bug tracker you guys used?? Thanks! Perceus does have a bug tracker here: http://www.perceus.org/portal/project/issues/perceus But I'll defer to the developers regarding where they would want the bugs filed. Cheers, Bernard From gmkurtzer at gmail.com Wed Jul 22 23:46:37 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Wed, 22 Jul 2009 23:46:37 -0700 Subject: [Warewulf] Bug with stateless provisioning of nash-based initrd with Perceus 1.6.0 In-Reply-To: References: <571f1a060907201942y303facb4y679a4673da88ced3@mail.gmail.com> Message-ID: <571f1a060907222346hbfcf053yf9435abcd3b5c8d7@mail.gmail.com> Thanks for this patch! Can you please check on SVN commit 2202 which changes the ordering of the creation of /nashinit and the call to /sbin/nash. Thanks! Greg On Wed, Jul 22, 2009 at 5:07 PM, Bernard Li wrote: > Hi Greg: > > On Mon, Jul 20, 2009 at 7:42 PM, Greg Kurtzer wrote: > >> That code wasnt tested yet, but the goal of it was to facilitate the >> switchroot onto the new tmpfs (instead of ramfs as you pointed out >> earlier). My guess is that it is still running on the ramfs which is >> what we were hoping to avoid. > > I have found a solution, please see the attached patch. > > When switchroot runs, it automatically umounts old /proc, /sys and > /dev, so I set those up to prevent error messages. ?Also, /sysroot > needs to be mounted separately from / (in this case I mounted it as > tmpfs). ?switchroot cannot be called by a subprocess, therefore the > only way to get it working is to call nash via exec. ?You would think > that "echo switchroot | exec /sbin/nash" would work, but it doesn't, > so the only way is to pass it a file with the switchroot command and > that's what I've done, and it works now. > > Special thanks to pjones in #anaconda for his help on figuring this out. > > Perhaps we should implement a small program to facilitate switchroot > instead of relying on nash... ?because right now this would work for > nash based systems (i.e. Fedora, Red Hat, CentOS, etc.) > > Oh and just to confirm that the new code puts the OS in tmpfs instead of ramfs: > > [/sbin/nash does not exist] > # df -h / > Filesystem ? ? ? ? ? ?Size ?Used Avail Use% Mounted on > none ? ? ? ? ? ? ? ? ? ? 0 ? ? 0 ? ? 0 ? - ?/ > > [/sbin/nash exists] > # df -h / > Filesystem ? ? ? ? ? ?Size ?Used Avail Use% Mounted on > none ? ? ? ? ? ? ? ? 1014M ?152M ?862M ?15% / > > Cheers, > > Bernard > > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf > > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From bernard at vanhpc.org Thu Jul 23 00:15:45 2009 From: bernard at vanhpc.org (Bernard Li) Date: Thu, 23 Jul 2009 00:15:45 -0700 Subject: [Warewulf] Bug with stateless provisioning of nash-based initrd with Perceus 1.6.0 In-Reply-To: <571f1a060907222346hbfcf053yf9435abcd3b5c8d7@mail.gmail.com> References: <571f1a060907201942y303facb4y679a4673da88ced3@mail.gmail.com> <571f1a060907222346hbfcf053yf9435abcd3b5c8d7@mail.gmail.com> Message-ID: Hi Greg: On Wed, Jul 22, 2009 at 11:46 PM, Greg Kurtzer wrote: > Can you please check on SVN commit 2202 which changes the ordering of > the creation of /nashinit and the call to /sbin/nash. Yeah, this is a bit confusing. The ordering doesn't really matter, as long as it's there in that block of code. You have to remember that /init is called by the kernel which is kexec'd, and that comes after the code block we are editing in 89-vnfsinit.sh. Perhaps we should put a comment there noting this gotcha. Anyway, I will test this out tomorrow -- thanks for checking in this patch (and the other one regarding the semi-colons). Cheers, Bernard From gmkurtzer at gmail.com Thu Jul 23 00:21:49 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Thu, 23 Jul 2009 00:21:49 -0700 Subject: [Warewulf] Bug with stateless provisioning of nash-based initrd with Perceus 1.6.0 In-Reply-To: References: <571f1a060907201942y303facb4y679a4673da88ced3@mail.gmail.com> <571f1a060907222346hbfcf053yf9435abcd3b5c8d7@mail.gmail.com> Message-ID: <571f1a060907230021j1d0927cei9833bdd3b56581f7@mail.gmail.com> Ahhh, right you are. As you can see, I am chugging through my TODO list. ;) hehe Thanks, Greg On Thu, Jul 23, 2009 at 12:15 AM, Bernard Li wrote: > Hi Greg: > > On Wed, Jul 22, 2009 at 11:46 PM, Greg Kurtzer wrote: > >> Can you please check on SVN commit 2202 which changes the ordering of >> the creation of /nashinit and the call to /sbin/nash. > > Yeah, this is a bit confusing. ?The ordering doesn't really matter, as > long as it's there in that block of code. ?You have to remember that > /init is called by the kernel which is kexec'd, and that comes after > the code block we are editing in 89-vnfsinit.sh. > > Perhaps we should put a comment there noting this gotcha. > > Anyway, I will test this out tomorrow -- thanks for checking in this > patch (and the other one regarding the semi-colons). > > Cheers, > > Bernard > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From gmkurtzer at gmail.com Thu Jul 23 00:23:55 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Thu, 23 Jul 2009 00:23:55 -0700 Subject: [Warewulf] perceus initiation In-Reply-To: References: Message-ID: <571f1a060907230023r42942b7fv75db3febbc9d3dc@mail.gmail.com> I think Bernard nailed the technical response. As far as the bug tracker goes, you can either post the bugs here or on the bug tracker. (although I personally I prefer emails. ;) Thanks, Greg On Wed, Jul 22, 2009 at 5:40 PM, Bernard Li wrote: > Hi Gary: > > On Wed, Jul 22, 2009 at 2:44 PM, Gary Leong wrote: > >> Is there a manual way to initialize Perceus?? I want to script out the >> installation of Perceus rather than have it prompt me for questions.? The >> first time you run perceus, it complains that perceus needs to be initalize >> even though I setup the MySQL database, and configured perceus.conf, >> defaults.conf, and dnsmasq.conf.? If I run the initialize scripts, it >> clobbers these files, and I have to do them all over. > > If memory serves, as long as /var/lib/perceus/database exists, then > Perceus would treat the system as initialized. ?It sounds like you > have already manually "initialized" Perceus -- does it not work after > your manual configuration? > > For me, I would prime the Perceus setup manually and then call these > commands as necessary: > > perceus configure nfs > perceus configure sshkeys > > Another option is to call perceus init by using an expect script. > > But I think what would be even better is for Perceus to support a way > of doing automated installation -- I would give my +1 behind that. > >> There are are some bugs with initialization scripts.? Should I list them >> here or is there a bug tracker you guys used?? Thanks! > > Perceus does have a bug tracker here: > > http://www.perceus.org/portal/project/issues/perceus > > But I'll defer to the developers regarding where they would want the bugs filed. > > Cheers, > > Bernard > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From gmkurtzer at gmail.com Thu Jul 23 00:25:57 2009 From: gmkurtzer at gmail.com (Greg Kurtzer) Date: Thu, 23 Jul 2009 00:25:57 -0700 Subject: [Warewulf] using ipaddr module to bring a nic up with no config. In-Reply-To: References: <571f1a060907211609i662452ebnb21a32de26b96782@mail.gmail.com> Message-ID: <571f1a060907230025y31481c5q90d5c2419cdd8809@mail.gmail.com> Thats interesting. Can you grab the latest ipaddr module from SVN (message me in IRC tomorrow if you need a pointer on this) and run by hand via: # NODENAME=n0000 perl ./50-ipaddr.pl and then check the output? Thanks! On Tue, Jul 21, 2009 at 5:31 PM, John Hanks wrote: > Sorry, I left out much pertinent information in my haste to click send > before rushing out of the office. > > [root at perceus ~]# perceus --version > Version=1.5.2 > Build=0.2092 > > with a line like yours I still get an odd ifcfg-eth1 file: > > DEVICE=eth1 > ONBOOT=yes > TYPE=ethernet > PROTO=static > BOOTPROTO=static > IPADDR=10.128.8.184 > NETMASK=255.255.0.0 > GATEWAY= > DEVICE=eth1 > ONBOOT=yes > TYPE=ethernet > MTU=9000 > > eth0 has a consistently odd pattern: > > DEVICE=eth0 > ONBOOT=yes > TYPE=ethernet > PROTO=static > BOOTPROTO=static > IPADDR=10.128.8.184 > NETMASK=255.255.0.0 > GATEWAY= > DEVICE=eth0 > ONBOOT=yes > TYPE=ethernet > PROTO=static > BOOTPROTO=static > IPADDR=10.128.8.184 > NETMASK=255.255.0.0 > GATEWAY= > PEERDNS=no > > as if both devices get configured twice. > > Thanks, > > jbh > > On Tue, Jul 21, 2009 at 7:09 PM, Greg Kurtzer wrote: >> Hi John! >> >> What version of Perceus is this? >> >> Have you tried removing the none? >> >> With this config: >> >> test* eth0(PEERDNS=no):[default]/[default] eth1(MTU=9000): >> >> I get: >> >> DEVICE=eth1 >> ONBOOT=yes >> TYPE=ethernet >> MTU=9000 >> >> There was some fixes that you may need that went into a SCM update: >> >> # perceus --version >> Version=1.5.2 >> Build=2072 >> >> >> Thanks, >> Greg >> >> >> On Tue, Jul 21, 2009 at 12:40 PM, John Hanks wrote: >>> I'm trying to bring up a nodes nic with no configuration so it can >>> then have one or more tagged VLANs configured on it. I've tried >>> numerous permutations but have been unsuccessful. Does anyone have a >>> working configuration for doing this with the ipaddr module? Basically >>> I'd like this to work as I expect it too >>> >>> node* eth0(PEERDNS=no):[default]/[default] eth1(MTU=9000):none >>> >>> and just bring eth1 up with no assigned address and with and mtu of >>> 9000. Instead I get this ifcfg-eth1 file which includes aspects of the >>> eth0 config: >>> >>> DEVICE=eth1 >>> ONBOOT=yes >>> TYPE=ethernet >>> PROTO=static >>> BOOTPROTO=static >>> IPADDR=10.128.8.185 >>> NETMASK=255.255.0.0 >>> GATEWAY= >>> DEVICE=eth1 >>> ONBOOT=yes >>> TYPE=ethernet >>> PROTO=static >>> BOOTPROTO=static >>> IPADDR=none >>> NETMASK= >>> GATEWAY= >>> MTU=9000 >>> >>> >>> Thanks, >>> >>> jbh >>> _______________________________________________ >>> Warewulf mailing list >>> Warewulf at caoslinux.org >>> http://lists.caosity.org/mailman/listinfo/warewulf >>> >> >> >> >> -- >> Greg Kurtzer >> http://www.infiscale.com/ >> http://www.perceus.org/ >> http://www.caoslinux.org/ >> _______________________________________________ >> Warewulf mailing list >> Warewulf at caoslinux.org >> http://lists.caosity.org/mailman/listinfo/warewulf >> > _______________________________________________ > Warewulf mailing list > Warewulf at caoslinux.org > http://lists.caosity.org/mailman/listinfo/warewulf > -- Greg Kurtzer http://www.infiscale.com/ http://www.perceus.org/ http://www.caoslinux.org/ From bernard at vanhpc.org Thu Jul 23 00:58:45 2009 From: bernard at vanhpc.org (Bernard Li) Date: Thu, 23 Jul 2009 00:58:45 -0700 Subject: [Warewulf] perceus initiation In-Reply-To: <571f1a060907230023r42942b7fv75db3febbc9d3dc@mail.gmail.com> References: <571f1a060907230023r42942b7fv75db3febbc9d3dc@mail.gmail.com> Message-ID: Hi Greg: On Thu, Jul 23, 2009 at 12:23 AM, Greg Kurtzer wrote: > I think Bernard nailed the technical response. As far as the bug > tracker goes, you can either post the bugs here or on the bug tracker. > (although I personally I prefer emails. ;) So what do you think about an unattended/automated install option for Perceus initialization? Something to be added to the TODO list? :-) Cheers, Bernard From bernard at vanhpc.org Thu Jul 23 11:52:17 2009 From: bernard at vanhpc.org (Bernard Li) Date: Thu, 23 Jul 2009 11:52:17 -0700 Subject: [Warewulf] Bug with stateless provisioning of nash-based initrd with Perceus 1.6.0 In-Reply-To: References: <571f1a060907201942y303facb4y679a4673da88ced3@mail.gmail.com> <571f1a060907222346hbfcf053yf9435abcd3b5c8d7@mail.gmail.com> Message-ID: Just tested r2203, and stateless provisioning to tmpfs works perfectly fine. Cheers, Bernard On Thu, Jul 23, 2009 at 12:15 AM, Bernard Li wrote: > Hi Greg: > > On Wed, Jul 22, 2009 at 11:46 PM, Greg Kurtzer wrote: > >> Can you please check on SVN commit 2202 which changes the ordering of >> the creation of /nashinit and the call to /sbin/nash. > > Yeah, this is a bit confusing. ?The ordering doesn't really matter, as > long as it's there in that block of code. ?You have to remember that > /init is called by the kernel which is kexec'd, and that comes after > the code block we are editing in 89-vnfsinit.sh. > > Perhaps we should put a comment there noting this gotcha. > > Anyway, I will test this out tomorrow -- thanks for checking in this > patch (and the other one regarding the semi-colons). > > Cheers, > > Bernard > From jbh at broadinstitute.org Thu Jul 23 18:40:04 2009 From: jbh at broadinstitute.org (John Hanks) Date: Thu, 23 Jul 2009 21:40:04 -0400 Subject: [Warewulf] using ipaddr module to bring a nic up with no config. In-Reply-To: <571f1a060907230025y31481c5q90d5c2419cdd8809@mail.gmail.com> References: <571f1a060907211609i662452ebnb21a32de26b96782@mail.gmail.com> <571f1a060907230025y31481c5q90d5c2419cdd8809@mail.gmail.com> Message-ID: On Thu, Jul 23, 2009 at 3:25 AM, Greg Kurtzer wrote: > Thats interesting. > > Can you grab the latest ipaddr module from SVN (message me in IRC > tomorrow if you need a pointer on this) and run by hand via: > > # NODENAME=n0000 perl ./50-ipaddr.pl > > and then check the output? > > Thanks! I grabbed the latest from trunk. Here's my ipaddr config: * eth0:[default]/[default] eth1:[default]/[default] node* eth0(PEERDNS=no):[default]/[default] eth1(MTU=9000): kvm-* eth0:[default]/[default] eth1(BOOTPROTO=none): Here's what happens when I run the script manually using one of my node names: [root at perceus nodescripts]# NODENAME=node1208 perl ./50-ipaddr.pl mkdir -p $DESTDIR/etc/sysconfig/network-scripts # This configuration has been automatically generated by the # 'ipaddr' Perceus module, and configured in: # /etc/perceus/modules/ipaddr echo 'DEVICE=eth0' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth0 echo 'ONBOOT=yes' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth0 echo 'TYPE=ethernet' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth0 echo 'PROTO=static' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth0 echo 'BOOTPROTO=static' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth0 echo 'IPADDR=10.128.8.184' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth0 echo 'NETMASK=255.255.0.0' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth0 echo 'GATEWAY=' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth0 mkdir -p $DESTDIR/etc/sysconfig/nics # This configuration has been automatically generated by the # 'ipaddr' Perceus module, and configured in: # /etc/perceus/modules/ipaddr echo 'DEVICE=eth0' >> $DESTDIR/etc/sysconfig/nics/./eth0 echo 'ONBOOT=yes' >> $DESTDIR/etc/sysconfig/nics/./eth0 echo 'TYPE=ethernet' >> $DESTDIR/etc/sysconfig/nics/./eth0 echo 'PROTO=static' >> $DESTDIR/etc/sysconfig/nics/./eth0 echo 'BOOTPROTO=static' >> $DESTDIR/etc/sysconfig/nics/./eth0 echo 'IPADDR=10.128.8.184' >> $DESTDIR/etc/sysconfig/nics/./eth0 echo 'NETMASK=255.255.0.0' >> $DESTDIR/etc/sysconfig/nics/./eth0 echo 'GATEWAY=' >> $DESTDIR/etc/sysconfig/nics/./eth0 mkdir -p $DESTDIR/etc/sysconfig/network-scripts # This configuration has been automatically generated by the # 'ipaddr' Perceus module, and configured in: # /etc/perceus/modules/ipaddr echo 'DEVICE=eth1' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth1 echo 'ONBOOT=yes' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth1 echo 'TYPE=ethernet' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth1 echo 'PROTO=static' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth1 echo 'BOOTPROTO=static' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth1 echo 'IPADDR=10.128.8.184' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth1 echo 'NETMASK=255.255.0.0' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth1 echo 'GATEWAY=' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth1 mkdir -p $DESTDIR/etc/sysconfig/nics # This configuration has been automatically generated by the # 'ipaddr' Perceus module, and configured in: # /etc/perceus/modules/ipaddr echo 'DEVICE=eth1' >> $DESTDIR/etc/sysconfig/nics/./eth1 echo 'ONBOOT=yes' >> $DESTDIR/etc/sysconfig/nics/./eth1 echo 'TYPE=ethernet' >> $DESTDIR/etc/sysconfig/nics/./eth1 echo 'PROTO=static' >> $DESTDIR/etc/sysconfig/nics/./eth1 echo 'BOOTPROTO=static' >> $DESTDIR/etc/sysconfig/nics/./eth1 echo 'IPADDR=10.128.8.184' >> $DESTDIR/etc/sysconfig/nics/./eth1 echo 'NETMASK=255.255.0.0' >> $DESTDIR/etc/sysconfig/nics/./eth1 echo 'GATEWAY=' >> $DESTDIR/etc/sysconfig/nics/./eth1 mkdir -p $DESTDIR/etc/sysconfig/network-scripts # This configuration has been automatically generated by the # 'ipaddr' Perceus module, and configured in: # /etc/perceus/modules/ipaddr echo 'DEVICE=eth0' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth0 echo 'ONBOOT=yes' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth0 echo 'TYPE=ethernet' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth0 echo 'PROTO=static' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth0 echo 'BOOTPROTO=static' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth0 echo 'IPADDR=10.128.8.184' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth0 echo 'NETMASK=255.255.0.0' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth0 echo 'GATEWAY=' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth0 echo 'PEERDNS=no' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth0 mkdir -p $DESTDIR/etc/sysconfig/nics # This configuration has been automatically generated by the # 'ipaddr' Perceus module, and configured in: # /etc/perceus/modules/ipaddr echo 'DEVICE=eth0' >> $DESTDIR/etc/sysconfig/nics/./eth0 echo 'ONBOOT=yes' >> $DESTDIR/etc/sysconfig/nics/./eth0 echo 'TYPE=ethernet' >> $DESTDIR/etc/sysconfig/nics/./eth0 echo 'PROTO=static' >> $DESTDIR/etc/sysconfig/nics/./eth0 echo 'BOOTPROTO=static' >> $DESTDIR/etc/sysconfig/nics/./eth0 echo 'IPADDR=10.128.8.184' >> $DESTDIR/etc/sysconfig/nics/./eth0 echo 'NETMASK=255.255.0.0' >> $DESTDIR/etc/sysconfig/nics/./eth0 echo 'GATEWAY=' >> $DESTDIR/etc/sysconfig/nics/./eth0 echo 'PEERDNS=no' >> $DESTDIR/etc/sysconfig/nics/./eth0 mkdir -p $DESTDIR/etc/sysconfig/network-scripts # This configuration has been automatically generated by the # 'ipaddr' Perceus module, and configured in: # /etc/perceus/modules/ipaddr echo 'DEVICE=eth1' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth1 echo 'ONBOOT=yes' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth1 echo 'TYPE=ethernet' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth1 echo 'MTU=9000' >> $DESTDIR/etc/sysconfig/network-scripts/ifcfg-eth1 mkdir -p $DESTDIR/etc/sysconfig/nics # This configuration has been automatically generated by the # 'ipaddr' Perceus module, and configured in: # /etc/perceus/modules/ipaddr echo 'DEVICE=eth1' >> $DESTDIR/etc/sysconfig/nics/./eth1 echo 'ONBOOT=yes' >> $DESTDIR/etc/sysconfig/nics/./eth1 echo 'TYPE=ethernet' >> $DESTDIR/etc/sysconfig/nics/./eth1 echo 'MTU=9000' >> $DESTDIR/etc/sysconfig/nics/./eth1 jbh From jbh at broadinstitute.org Thu Jul 23 19:13:56 2009 From: jbh at broadinstitute.org (John Hanks) Date: Thu, 23 Jul 2009 22:13:56 -0400 Subject: [Warewulf] using ipaddr module to bring a nic up with no config. In-Reply-To: References: <571f1a060907211609i662452ebnb21a32de26b96782@mail.gmail.com> <571f1a060907230025y31481c5q90d5c2419cdd8809@mail.gmail.com> Message-ID: On Thu, Jul 23, 2009 at 9:40 PM, John Hanks wrote: > * eth0:[default]/[default] eth1:[default]/[default] > node* eth0(PEERDNS=no):[default]/[default] eth1(MTU=9000): I tried a few more permutations and it appears that I get a config for every matching line in the ipaddr config. If the first echo in each used > instead of >> the correct (last) config would win or if only the most accurate match produced a config it would work as expected. For now I can easily workaround by commenting out the generic * line so that only one line matches for a given node. jbh From stefano.bridi at gmail.com Thu Jul 30 07:42:06 2009 From: stefano.bridi at gmail.com (Stefano Bridi) Date: Thu, 30 Jul 2009 16:42:06 +0200 Subject: [Warewulf] perceus 1.5.3 on centos 5.3 bugfix proposal Message-ID: Hi all, I'm playing with a small test cluster with a fresh install of centos 5.3 + centosplus + contrib + rpmforge repos where the nodes wont boot. After some debugging started a line in from /var/log/messages repeated in a loop... perceus[25106]: WARN: Could not resolve node's ID/MAC addr from 10.10.255.0, skipping. I found that the problem was in perceusd at the line 364: $NodeID = uc(Net::ARP::arp_lookup("", $ipaddr)); The function fail to get the MAC/NodeID I solved with this line: $NodeID = uc(Net::ARP::arp_lookup($eth, $ipaddr)); In this way the parameter "master network device" became mandatory but if I understand correctly the source and the docs It is already mandatory... bye stef From bernard at vanhpc.org Fri Jul 31 15:28:56 2009 From: bernard at vanhpc.org (Bernard Li) Date: Fri, 31 Jul 2009 15:28:56 -0700 Subject: [Warewulf] [PATCH] Fix initramfs mtime for kernel < 2.6.28 Message-ID: Dear all: If you have booted a Perceus node with kernel < 2.6.28 (eg. EL5 comes with 2.6.18), you will notice that the timestamp of files when you do a `ls -l` is the time of provision time, and not that of when the files were last modified. This is due to the fact that kernels prior to 2.6.28 when they decompress the initramfs (for Perceus this is vnfs.img, i.e. your VNFS image), they do not actually retain the mtimes. The attached patch works around this issue. Basically it adds a new config option VNFSTIME and when enabled, will: - create a vnfs.lst in your $VNFSDIR consisting of files in your image, modification time and type (symlink, file, directory, etc.) - download this vnfs.lst when the node boots - during init, it will go through your filesystem and modify the timestamp of each non-symlink file/dir For an image that has ~5000 files, it takes an additional ~7 seconds to boot. The code basically sorts the file list by mtime, and would `touch` files grouped by mtime using xargs. Fun facts: - Using Perl, it took ~0.455 seconds - Using Python, it took ~0.8 seconds - Without the time sort and xargs, it would take ~20 seconds I opted to use system commands to make it more portable, as not all capsules would include Perl or Python. It would be great to be able to reduce the time further, but I think it is acceptable for now. Comments, suggestions are welcome. Thanks, Bernard -------------- next part -------------- A non-text attachment was scrubbed... Name: chroot2stateless.sh.patch Type: text/x-patch Size: 3550 bytes Desc: not available Url : http://altruistic.infiscale.org/pipermail/perceus/attachments/20090731/851cf63c/attachment.bin