Re: VCL2.2 + xCAT2.5 on bladecenter
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 John, Do you have IMAGELIBINSTALLDIR set to /opt or to /opt/image/x86? It needs to be /opt. One thing to try is changing usepartimageng to 1 on line 147. That will switch to using partimageng instead of partimage and tell us if it is something specific to partimage. The next step will be to find the partimage logs, which is a little more complicated. Feel free to hop on IRC to debug a little faster: #asfvcl on freenode. Josh On Friday April 01, 2011, John Ma wrote: > Josh, > > I added the debugging code. It turned out NFS looks fine. I also manually > verified the NFS share: > > [root@blade14 ~]# mkdir nfstest > [root@blade14 ~]# mount 172.20.0.1:/opt/image/x86 nfstest/ > [root@blade14 ~]# cd nfstest/ > [root@blade14 nfstest]# mkdir writetest > > > > How to debug the partimage save operation? > > Thanks, > John Ma > Marist College > > > > > From: Josh Thompson > To: vcl-user@incubator.apache.org > Date: 04/01/2011 09:06 AM > Subject:Re: VCL2.2 + xCAT2.5 on bladecenter > > > > - gpg control packet > John, > > The partimageng postscript mounts an image store via NFS at /install. The > NFS > server and path are specified in the xCAT site table as IMAGELIBSERVER and > > IMAGELIBINSTALLDIR. More info about this part is at the bottom of our > wiki > page explaining how to add partimage support to xCAT. > > Do you have your image store exported read/write via NFS and available to > the > client nodes? > > As a test, you could modify the partimageng script to output more > debugging > info. You could modify the mount command on line 144 to be: > > logger -t xcat "Attempting to mount image store: > $IMAGELIBSERVER:$IMAGELIBINSTALLDIR" > if ! mount -o nfsvers=3,tcp,nolock,rw $IMAGELIBSERVER:$IMAGELIBINSTALLDIR > /install; then > echo "CRITICAL ERROR: Failed to mount image store at > $IMAGELIBSERVER:$IMAGELIBINSTALLDIR; unable to save image" > logger -t xcat "CRITICAL ERROR: Failed to mount image store at > $IMAGELIBSERVER:$IMAGELIBINSTALLDIR; unable to save image" > sleep 3 > exit 1 > fi > > Josh > > On Thursday March 31, 2011, John Ma wrote: > > Josh, > > > > Thanks again for the help. I configured an anonymous ftp share of > > > > /install, and it passed the previous error. Now I am at here: > > Mar 30 21:06:53 blade08 blade08 xcat: running partimage -z1 -f3 -odbc > > > > save /dev/sda1 /install/image/x86/centos5image-blade08mar2466-v0.gz > > > > Mar 30 21:06:58 blade08 blade08 xcat: partimage exited with a non-zero > > > > status, failing > > > > Mar 30 21:06:58 blade08 blade08 xcat: partimage-ng failed with exit > > code > > > 1 > > > > Mar 30 21:06:58 blade08 blade08 init: rc3 main process (1166) killed > > by > > > TERM signal > > > > Blade08 then rebooted itself and loop again. partimage's save location > > /install/image/x86/.. doesn't seem right to me, but how to configure it > > to > > > use nfs? > > See the attached log file for more details, ( the clock on blade08 is > > off > > > or maybe UTC) > > > > > > Thanks, > > John > > > > > > > > > > > > > > From: Josh Thompson > > To: vcl-user@incubator.apache.org > > Date: 03/31/2011 03:29 PM > > Subject:Re: VCL2.2 + xCAT2.5 on bladecenter > > > > > > > > - gpg control packet > > John, > > > > Sorry to take so long to get back to you. > > > > I didn't even realize this until digging through xcatdsklspost, but your > > management node needs to be sharing out /install via ftp. I'm assuming > > xcat > > sets this up because I don't remember setting that up manually. The > > following > > line is from xcatdsklspost: > > > > wget -l inf -N -r --waitretry=10 --random-wait --retry-connrefused -t 0 > > -T > > > 60 > > ftp://$SIP/postscripts 2> /tmp/wget.log > > > > $SIP is obtained earlier in the script from some dhcp information. > > > > The next line is where your screenshot shows the first error: > > > > mv $SIP/postscripts/* /xcatpost; > > > > The wget command should try forever until it downloads everything under > > ftp://$SIP/postscripts. The fact that you are getting past wget, but > > the > > > move > > is failing for $SIP/postscripts/* makes me think you don't have anything &g
Re: VCL2.2 + xCAT2.5 on bladecenter
Josh, I added the debugging code. It turned out NFS looks fine. I also manually verified the NFS share: [root@blade14 ~]# mkdir nfstest [root@blade14 ~]# mount 172.20.0.1:/opt/image/x86 nfstest/ [root@blade14 ~]# cd nfstest/ [root@blade14 nfstest]# mkdir writetest How to debug the partimage save operation? Thanks, John Ma Marist College From: Josh Thompson To: vcl-user@incubator.apache.org Date: 04/01/2011 09:06 AM Subject:Re: VCL2.2 + xCAT2.5 on bladecenter -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 John, The partimageng postscript mounts an image store via NFS at /install. The NFS server and path are specified in the xCAT site table as IMAGELIBSERVER and IMAGELIBINSTALLDIR. More info about this part is at the bottom of our wiki page explaining how to add partimage support to xCAT. Do you have your image store exported read/write via NFS and available to the client nodes? As a test, you could modify the partimageng script to output more debugging info. You could modify the mount command on line 144 to be: logger -t xcat "Attempting to mount image store: $IMAGELIBSERVER:$IMAGELIBINSTALLDIR" if ! mount -o nfsvers=3,tcp,nolock,rw $IMAGELIBSERVER:$IMAGELIBINSTALLDIR /install; then echo "CRITICAL ERROR: Failed to mount image store at $IMAGELIBSERVER:$IMAGELIBINSTALLDIR; unable to save image" logger -t xcat "CRITICAL ERROR: Failed to mount image store at $IMAGELIBSERVER:$IMAGELIBINSTALLDIR; unable to save image" sleep 3 exit 1 fi Josh On Thursday March 31, 2011, John Ma wrote: > Josh, > > Thanks again for the help. I configured an anonymous ftp share of > /install, and it passed the previous error. Now I am at here: > Mar 30 21:06:53 blade08 blade08 xcat: running partimage -z1 -f3 -odbc > save /dev/sda1 /install/image/x86/centos5image-blade08mar2466-v0.gz > Mar 30 21:06:58 blade08 blade08 xcat: partimage exited with a non-zero > status, failing > Mar 30 21:06:58 blade08 blade08 xcat: partimage-ng failed with exit code > 1 > Mar 30 21:06:58 blade08 blade08 init: rc3 main process (1166) killed by > TERM signal > > Blade08 then rebooted itself and loop again. partimage's save location > /install/image/x86/.. doesn't seem right to me, but how to configure it to > use nfs? > See the attached log file for more details, ( the clock on blade08 is off > or maybe UTC) > > > Thanks, > John > > > > > > > From: Josh Thompson > To: vcl-user@incubator.apache.org > Date: 03/31/2011 03:29 PM > Subject:Re: VCL2.2 + xCAT2.5 on bladecenter > > > > - gpg control packet > John, > > Sorry to take so long to get back to you. > > I didn't even realize this until digging through xcatdsklspost, but your > management node needs to be sharing out /install via ftp. I'm assuming > xcat > sets this up because I don't remember setting that up manually. The > following > line is from xcatdsklspost: > > wget -l inf -N -r --waitretry=10 --random-wait --retry-connrefused -t 0 -T > 60 > ftp://$SIP/postscripts 2> /tmp/wget.log > > $SIP is obtained earlier in the script from some dhcp information. > > The next line is where your screenshot shows the first error: > > mv $SIP/postscripts/* /xcatpost; > > The wget command should try forever until it downloads everything under > ftp://$SIP/postscripts. The fact that you are getting past wget, but the > move > is failing for $SIP/postscripts/* makes me think you don't have anything > under > ftp://$SIP/postscripts. Can you try using a normal ftp client to browse > ftp://172.20.101.140/postscripts? It may be that the ftp server is > sharing > out the wrong directory. > > Josh > > On Friday March 25, 2011, John Ma wrote: > > Josh, > > > > I made some progress, but stuck again. This time at the reboot of the > > machine being captured. The machine apparently cannot find postscripts. > > Any idea about how to fix it or what to try next? > > > > I placed partimageng in /install/postscripts on our VCL (web, db, and > > mgt > > > code) server - Blade14 (172.20.101.140). The machine being captured is > > blade08 (172.20.101.80). > > > > Here is the screenshot: > > > > Here is the pxe boot config file: > > [root@blade14 ~]# cat /tftpboot/pxelinux.cfg/blade08 > > #image image-x86-centos5image-blade08mar2466-v0 > > DEFAULT xCAT > > LABEL xCAT > > > > KERNEL xcat/image/x86/vmlinuz > > APPEND initrd=xcat/image/x86/initrd.img > > > > imgurl=http://blade14//install/image/x86/installer_files/rootimg.gz > > image=/install/image/x86/centos5image-blade08mar2466-v
Re: VCL2.2 + xCAT2.5 on bladecenter
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 John, The partimageng postscript mounts an image store via NFS at /install. The NFS server and path are specified in the xCAT site table as IMAGELIBSERVER and IMAGELIBINSTALLDIR. More info about this part is at the bottom of our wiki page explaining how to add partimage support to xCAT. Do you have your image store exported read/write via NFS and available to the client nodes? As a test, you could modify the partimageng script to output more debugging info. You could modify the mount command on line 144 to be: logger -t xcat "Attempting to mount image store: $IMAGELIBSERVER:$IMAGELIBINSTALLDIR" if ! mount -o nfsvers=3,tcp,nolock,rw $IMAGELIBSERVER:$IMAGELIBINSTALLDIR /install; then echo "CRITICAL ERROR: Failed to mount image store at $IMAGELIBSERVER:$IMAGELIBINSTALLDIR; unable to save image" logger -t xcat "CRITICAL ERROR: Failed to mount image store at $IMAGELIBSERVER:$IMAGELIBINSTALLDIR; unable to save image" sleep 3 exit 1 fi Josh On Thursday March 31, 2011, John Ma wrote: > Josh, > > Thanks again for the help. I configured an anonymous ftp share of > /install, and it passed the previous error. Now I am at here: > Mar 30 21:06:53 blade08 blade08 xcat: running partimage -z1 -f3 -odbc > save /dev/sda1 /install/image/x86/centos5image-blade08mar2466-v0.gz > Mar 30 21:06:58 blade08 blade08 xcat: partimage exited with a non-zero > status, failing > Mar 30 21:06:58 blade08 blade08 xcat: partimage-ng failed with exit code > 1 > Mar 30 21:06:58 blade08 blade08 init: rc3 main process (1166) killed by > TERM signal > > Blade08 then rebooted itself and loop again. partimage's save location > /install/image/x86/.. doesn't seem right to me, but how to configure it to > use nfs? > See the attached log file for more details, ( the clock on blade08 is off > or maybe UTC) > > > Thanks, > John > > > > > > > From: Josh Thompson > To: vcl-user@incubator.apache.org > Date: 03/31/2011 03:29 PM > Subject:Re: VCL2.2 + xCAT2.5 on bladecenter > > > > - gpg control packet > John, > > Sorry to take so long to get back to you. > > I didn't even realize this until digging through xcatdsklspost, but your > management node needs to be sharing out /install via ftp. I'm assuming > xcat > sets this up because I don't remember setting that up manually. The > following > line is from xcatdsklspost: > > wget -l inf -N -r --waitretry=10 --random-wait --retry-connrefused -t 0 -T > 60 > ftp://$SIP/postscripts 2> /tmp/wget.log > > $SIP is obtained earlier in the script from some dhcp information. > > The next line is where your screenshot shows the first error: > > mv $SIP/postscripts/* /xcatpost; > > The wget command should try forever until it downloads everything under > ftp://$SIP/postscripts. The fact that you are getting past wget, but the > move > is failing for $SIP/postscripts/* makes me think you don't have anything > under > ftp://$SIP/postscripts. Can you try using a normal ftp client to browse > ftp://172.20.101.140/postscripts? It may be that the ftp server is > sharing > out the wrong directory. > > Josh > > On Friday March 25, 2011, John Ma wrote: > > Josh, > > > > I made some progress, but stuck again. This time at the reboot of the > > machine being captured. The machine apparently cannot find postscripts. > > Any idea about how to fix it or what to try next? > > > > I placed partimageng in /install/postscripts on our VCL (web, db, and > > mgt > > > code) server - Blade14 (172.20.101.140). The machine being captured is > > blade08 (172.20.101.80). > > > > Here is the screenshot: > > > > Here is the pxe boot config file: > > [root@blade14 ~]# cat /tftpboot/pxelinux.cfg/blade08 > > #image image-x86-centos5image-blade08mar2466-v0 > > DEFAULT xCAT > > LABEL xCAT > > > > KERNEL xcat/image/x86/vmlinuz > > APPEND initrd=xcat/image/x86/initrd.img > > > > imgurl=http://blade14//install/image/x86/installer_files/rootimg.gz > > image=/install/image/x86/centos5image-blade08mar2466-v0.img blocks=512 > > action=save installnic=eth0 reboot noipv6 > > > > IPAPPEND 2 > > > > [root@blade14 ~]# > > > > Thanks, > > John Ma > > Marist College > > > > > > > > > > From: Josh Thompson > > To: vcl-user@incubator.apache.org > > Date: 03/22/2011 12:15 PM > > Subject:Re: VCL2.2 + xCAT2.5 on bladecenter - -- - --- Josh Thompson VCL Developer North Carolina State University my GPG/PGP key can be found at pgp.mit.edu -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.16 (GNU/Linux) iEYEARECAAYFAk2VzaAACgkQV/LQcNdtPQPPfwCfZF5WlXYUnvLAV6XXiPG4ENQe k7MAnAgiDIaXKzr8Lr9dClRuVGp6peaK =MuU1 -END PGP SIGNATURE-
Re: VCL2.2 + xCAT2.5 on bladecenter
Josh, Thanks again for the help. I configured an anonymous ftp share of /install, and it passed the previous error. Now I am at here: Mar 30 21:06:53 blade08 blade08 xcat: running partimage -z1 -f3 -odbc save /dev/sda1 /install/image/x86/centos5image-blade08mar2466-v0.gz Mar 30 21:06:58 blade08 blade08 xcat: partimage exited with a non-zero status, failing Mar 30 21:06:58 blade08 blade08 xcat: partimage-ng failed with exit code 1 Mar 30 21:06:58 blade08 blade08 init: rc3 main process (1166) killed by TERM signal Blade08 then rebooted itself and loop again. partimage's save location /install/image/x86/.. doesn't seem right to me, but how to configure it to use nfs? See the attached log file for more details, ( the clock on blade08 is off or maybe UTC) Thanks, John From: Josh Thompson To: vcl-user@incubator.apache.org Date: 03/31/2011 03:29 PM Subject: Re: VCL2.2 + xCAT2.5 on bladecenter -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 John, Sorry to take so long to get back to you. I didn't even realize this until digging through xcatdsklspost, but your management node needs to be sharing out /install via ftp. I'm assuming xcat sets this up because I don't remember setting that up manually. The following line is from xcatdsklspost: wget -l inf -N -r --waitretry=10 --random-wait --retry-connrefused -t 0 -T 60 ftp://$SIP/postscripts 2> /tmp/wget.log $SIP is obtained earlier in the script from some dhcp information. The next line is where your screenshot shows the first error: mv $SIP/postscripts/* /xcatpost; The wget command should try forever until it downloads everything under ftp://$SIP/postscripts. The fact that you are getting past wget, but the move is failing for $SIP/postscripts/* makes me think you don't have anything under ftp://$SIP/postscripts. Can you try using a normal ftp client to browse ftp://172.20.101.140/postscripts? It may be that the ftp server is sharing out the wrong directory. Josh On Friday March 25, 2011, John Ma wrote: > Josh, > > I made some progress, but stuck again. This time at the reboot of the > machine being captured. The machine apparently cannot find postscripts. > Any idea about how to fix it or what to try next? > > I placed partimageng in /install/postscripts on our VCL (web, db, and mgt > code) server - Blade14 (172.20.101.140). The machine being captured is > blade08 (172.20.101.80). > > Here is the screenshot: > > Here is the pxe boot config file: > [root@blade14 ~]# cat /tftpboot/pxelinux.cfg/blade08 > #image image-x86-centos5image-blade08mar2466-v0 > DEFAULT xCAT > LABEL xCAT > KERNEL xcat/image/x86/vmlinuz > APPEND initrd=xcat/image/x86/initrd.img > imgurl=http://blade14//install/image/x86/installer_files/rootimg.gz > image=/install/image/x86/centos5image-blade08mar2466-v0.img blocks=512 > action=save installnic=eth0 reboot noipv6 > IPAPPEND 2 > [root@blade14 ~]# > > Thanks, > John Ma > Marist College > > > > > From: Josh Thompson > To: vcl-user@incubator.apache.org > Date: 03/22/2011 12:15 PM > Subject:Re: VCL2.2 + xCAT2.5 on bladecenter > > > - -- - --- Josh Thompson VCL Developer North Carolina State University my GPG/PGP key can be found at pgp.mit.edu -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.16 (GNU/Linux) iEYEARECAAYFAk2U1e0ACgkQV/LQcNdtPQOKWACeMKxRbAc6eQlzg5AsGA7Y0i1F 4YIAn1mwoaOfelY2vsFhTXr9Yw2NpOEf =sW7U -END PGP SIGNATURE- message Description: Binary data
Re: VCL2.2 + xCAT2.5 on bladecenter
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 John, Sorry to take so long to get back to you. I didn't even realize this until digging through xcatdsklspost, but your management node needs to be sharing out /install via ftp. I'm assuming xcat sets this up because I don't remember setting that up manually. The following line is from xcatdsklspost: wget -l inf -N -r --waitretry=10 --random-wait --retry-connrefused -t 0 -T 60 ftp://$SIP/postscripts 2> /tmp/wget.log $SIP is obtained earlier in the script from some dhcp information. The next line is where your screenshot shows the first error: mv $SIP/postscripts/* /xcatpost; The wget command should try forever until it downloads everything under ftp://$SIP/postscripts. The fact that you are getting past wget, but the move is failing for $SIP/postscripts/* makes me think you don't have anything under ftp://$SIP/postscripts. Can you try using a normal ftp client to browse ftp://172.20.101.140/postscripts? It may be that the ftp server is sharing out the wrong directory. Josh On Friday March 25, 2011, John Ma wrote: > Josh, > > I made some progress, but stuck again. This time at the reboot of the > machine being captured. The machine apparently cannot find postscripts. > Any idea about how to fix it or what to try next? > > I placed partimageng in /install/postscripts on our VCL (web, db, and mgt > code) server - Blade14 (172.20.101.140). The machine being captured is > blade08 (172.20.101.80). > > Here is the screenshot: > > Here is the pxe boot config file: > [root@blade14 ~]# cat /tftpboot/pxelinux.cfg/blade08 > #image image-x86-centos5image-blade08mar2466-v0 > DEFAULT xCAT > LABEL xCAT > KERNEL xcat/image/x86/vmlinuz > APPEND initrd=xcat/image/x86/initrd.img > imgurl=http://blade14//install/image/x86/installer_files/rootimg.gz > image=/install/image/x86/centos5image-blade08mar2466-v0.img blocks=512 > action=save installnic=eth0 reboot noipv6 > IPAPPEND 2 > [root@blade14 ~]# > > Thanks, > John Ma > Marist College > > > > > From: Josh Thompson > To: vcl-user@incubator.apache.org > Date: 03/22/2011 12:15 PM > Subject:Re: VCL2.2 + xCAT2.5 on bladecenter > > > - -- - --- Josh Thompson VCL Developer North Carolina State University my GPG/PGP key can be found at pgp.mit.edu -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.16 (GNU/Linux) iEYEARECAAYFAk2U1e0ACgkQV/LQcNdtPQOKWACeMKxRbAc6eQlzg5AsGA7Y0i1F 4YIAn1mwoaOfelY2vsFhTXr9Yw2NpOEf =sW7U -END PGP SIGNATURE-
Re: VCL2.2 + xCAT2.5 on bladecenter
Hi, I am over this hurdle after reboot the management node. Lesson Learned: xcatd will pickup newly inserted plugin at startup time. Thanks, John Ma Marist College From: John Ma To: vcl-user@incubator.apache.org Date: 03/23/2011 10:45 AM Subject:Re: VCL2.2 + xCAT2.5 on bladecenter Hi, Just to clarify that I didn't attempt to rebuild the three stateless images, only to use them as is. So I probably don't need to worry about choosing partimage or partimage-ng. I still stuck at 'Unable to identify plugin for this command, check relevant tables: nodetype.os', any help/hint/suggestion is appreciated. Thanks, John Ma Marist College From:John Ma To:vcl-user@incubator.apache.org Date:03/22/2011 03:55 PM Subject: Re: VCL2.2 + xCAT2.5 on bladecenter Josh, Thank you for the detailed instruction, I gave it a try to setup xCAT2.5/imaging by following https://cwiki.apache.org/VCL/adding-support-for-partimage-and-partimage-ng-to-xcat-2x-unofficial.html . I cannot find partimage-ng anywhere, so I have partimage-0.6.9 compiled and installed, but it doesn't seem working. I got: [root@blade14 ~]# nodeset blade08 image blade08: Error: Unable to identify plugin for this command, check relevant tables: nodetype.os I have to handcraft several xcat tables, among them nodetype and osimage are the ones I have little confidence. Could you dump some relevant records in those two tables to illustrate your setup? also, do I need to change anything to switch from partimage-ng to partimage? Thanks, John Ma Marist College Poughkeepsie, NY From:Josh Thompson To:vcl-user@incubator.apache.org Date:03/22/2011 12:15 PM Subject: Re: VCL2.2 + xCAT2.5 on bladecenter -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 John, The stateless parts are not actually black boxes. They are gziped cpio files and are typical linux initrd and root filesystem images. You can do the following to extract them: mkdir /tmp/stateless cd /tmp/stateless mkdir initrd mkdir rootimg wget http://people.apache.org/~jfthomps/partimageng_xCAT2x/initrd.img wget http://people.apache.org/~jfthomps/partimageng_xCAT2x/rootimg.gz cd initrd zcat ../initrd.img | cpio --extract cd ../rootimg zcat ../rootimg.gz | cpio --extract # if you do this last command as a non- root user, you will get one error about not being able to create dev/null which is okay to ignore unless you want to modify and recreate the image You would actually need to go through this process if you have hardware that is not supported in these images. I only had the hardware at NCSU to develop on; so, they don't have a very big range of hardware support. If you decide to go ahead with using this stuff for partimage and need support for different hardware, I can guide you through adding it in to initrd and rootimg. If you wanted to modify and recreate the images after doing the above, you would do the following: cd /tmp/stateless/initrd find . | cpio -oc | gzip -c -9 > ../initrd-new.img cd /tmp/stateless/rootimg find . | cpio -oc | gzip -c -9 > ../rootimg-new.gz The two images really don't do much. Most of the work is done by the partimageng postscript. They really just provide a small install of linux on which the other stuff can run. As long as NCSU is using bare metal images, we'll still need a way to use xCAT and partimage to deploy images, and I don't see us moving completely away from bare metal images for many years. I have been told that more recent versions of xCAT 2 do have partimage support, but that you have to download that part separately. I have not looked in to this, but if it would be easy to switch VCL over to using that, it's possible we may do so in the future. Josh On Tuesday March 22, 2011, John Ma wrote: > My concerns to the pre-build images are > 1. Security - It's a black box. Will need our security officer's > approval for production use. > 2. Support - Will it still be available when we need to expand to other > hardware platform or next release of xCAT/partimage? > > John Ma > Marist College > > > > > From: Aaron Peeler > To: vcl-user@incubator.apache.org > Date: 03/22/2011 10:32 AM > Subject:Re: VCL2.2 + xCAT2.5 on bladecenter > > > > We like partimage-ng. > > But, we still use both partimage and partimage-ng. Our case is a little > unique, we started using xCAT1.3 with partimage, then added local > support ourselves because xcat 2.0 did not include imaging. > > Can you expand on what you feel uncomfortable about? Is it building them > or modifying, you should be able to mount them and view the contents. > > Aaron > > On 3/21/11 1:53 PM, John Ma wrote: > > Hi Aaron, > > > > I couldn
Re: VCL2.2 + xCAT2.5 on bladecenter
Hi, Just to clarify that I didn't attempt to rebuild the three stateless images, only to use them as is. So I probably don't need to worry about choosing partimage or partimage-ng. I still stuck at 'Unable to identify plugin for this command, check relevant tables: nodetype.os', any help/hint/suggestion is appreciated. Thanks, John Ma Marist College From: John Ma To: vcl-user@incubator.apache.org Date: 03/22/2011 03:55 PM Subject: Re: VCL2.2 + xCAT2.5 on bladecenter Josh, Thank you for the detailed instruction, I gave it a try to setup xCAT2.5/imaging by following https://cwiki.apache.org/VCL/adding-support-for-partimage-and-partimage-ng-to-xcat-2x-unofficial.html . I cannot find partimage-ng anywhere, so I have partimage-0.6.9 compiled and installed, but it doesn't seem working. I got: [root@blade14 ~]# nodeset blade08 image blade08: Error: Unable to identify plugin for this command, check relevant tables: nodetype.os I have to handcraft several xcat tables, among them nodetype and osimage are the ones I have little confidence. Could you dump some relevant records in those two tables to illustrate your setup? also, do I need to change anything to switch from partimage-ng to partimage? Thanks, John Ma Marist College Poughkeepsie, NY From:Josh Thompson To:vcl-user@incubator.apache.org Date:03/22/2011 12:15 PM Subject: Re: VCL2.2 + xCAT2.5 on bladecenter -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 John, The stateless parts are not actually black boxes. They are gziped cpio files and are typical linux initrd and root filesystem images. You can do the following to extract them: mkdir /tmp/stateless cd /tmp/stateless mkdir initrd mkdir rootimg wget http://people.apache.org/~jfthomps/partimageng_xCAT2x/initrd.img wget http://people.apache.org/~jfthomps/partimageng_xCAT2x/rootimg.gz cd initrd zcat ../initrd.img | cpio --extract cd ../rootimg zcat ../rootimg.gz | cpio --extract # if you do this last command as a non- root user, you will get one error about not being able to create dev/null which is okay to ignore unless you want to modify and recreate the image You would actually need to go through this process if you have hardware that is not supported in these images. I only had the hardware at NCSU to develop on; so, they don't have a very big range of hardware support. If you decide to go ahead with using this stuff for partimage and need support for different hardware, I can guide you through adding it in to initrd and rootimg. If you wanted to modify and recreate the images after doing the above, you would do the following: cd /tmp/stateless/initrd find . | cpio -oc | gzip -c -9 > ../initrd-new.img cd /tmp/stateless/rootimg find . | cpio -oc | gzip -c -9 > ../rootimg-new.gz The two images really don't do much. Most of the work is done by the partimageng postscript. They really just provide a small install of linux on which the other stuff can run. As long as NCSU is using bare metal images, we'll still need a way to use xCAT and partimage to deploy images, and I don't see us moving completely away from bare metal images for many years. I have been told that more recent versions of xCAT 2 do have partimage support, but that you have to download that part separately. I have not looked in to this, but if it would be easy to switch VCL over to using that, it's possible we may do so in the future. Josh On Tuesday March 22, 2011, John Ma wrote: > My concerns to the pre-build images are > 1. Security - It's a black box. Will need our security officer's > approval for production use. > 2. Support - Will it still be available when we need to expand to other > hardware platform or next release of xCAT/partimage? > > John Ma > Marist College > > > > > From: Aaron Peeler > To: vcl-user@incubator.apache.org > Date: 03/22/2011 10:32 AM > Subject:Re: VCL2.2 + xCAT2.5 on bladecenter > > > > We like partimage-ng. > > But, we still use both partimage and partimage-ng. Our case is a little > unique, we started using xCAT1.3 with partimage, then added local > support ourselves because xcat 2.0 did not include imaging. > > Can you expand on what you feel uncomfortable about? Is it building them > or modifying, you should be able to mount them and view the contents. > > Aaron > > On 3/21/11 1:53 PM, John Ma wrote: > > Hi Aaron, > > > > I couldn't even finish configuring xCAT2.5 to capture image. What > > imaging module is in your setup, partimage, partimage-ng, systemimager > > or other? I have this resource > > https://cwiki.apache.org/VCL/adding-support-for-partimage-and-partimage-ng- > to-xcat-2x-unofficial.html , > > > but feel unco
Re: VCL2.2 + xCAT2.5 on bladecenter
Josh, Thank you for the detailed instruction, I gave it a try to setup xCAT2.5/imaging by following https://cwiki.apache.org/VCL/adding-support-for-partimage-and-partimage-ng-to-xcat-2x-unofficial.html . I cannot find partimage-ng anywhere, so I have partimage-0.6.9 compiled and installed, but it doesn't seem working. I got: [root@blade14 ~]# nodeset blade08 image blade08: Error: Unable to identify plugin for this command, check relevant tables: nodetype.os I have to handcraft several xcat tables, among them nodetype and osimage are the ones I have little confidence. Could you dump some relevant records in those two tables to illustrate your setup? also, do I need to change anything to switch from partimage-ng to partimage? Thanks, John Ma Marist College Poughkeepsie, NY From: Josh Thompson To: vcl-user@incubator.apache.org Date: 03/22/2011 12:15 PM Subject: Re: VCL2.2 + xCAT2.5 on bladecenter -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 John, The stateless parts are not actually black boxes. They are gziped cpio files and are typical linux initrd and root filesystem images. You can do the following to extract them: mkdir /tmp/stateless cd /tmp/stateless mkdir initrd mkdir rootimg wget http://people.apache.org/~jfthomps/partimageng_xCAT2x/initrd.img wget http://people.apache.org/~jfthomps/partimageng_xCAT2x/rootimg.gz cd initrd zcat ../initrd.img | cpio --extract cd ../rootimg zcat ../rootimg.gz | cpio --extract # if you do this last command as a non- root user, you will get one error about not being able to create dev/null which is okay to ignore unless you want to modify and recreate the image You would actually need to go through this process if you have hardware that is not supported in these images. I only had the hardware at NCSU to develop on; so, they don't have a very big range of hardware support. If you decide to go ahead with using this stuff for partimage and need support for different hardware, I can guide you through adding it in to initrd and rootimg. If you wanted to modify and recreate the images after doing the above, you would do the following: cd /tmp/stateless/initrd find . | cpio -oc | gzip -c -9 > ../initrd-new.img cd /tmp/stateless/rootimg find . | cpio -oc | gzip -c -9 > ../rootimg-new.gz The two images really don't do much. Most of the work is done by the partimageng postscript. They really just provide a small install of linux on which the other stuff can run. As long as NCSU is using bare metal images, we'll still need a way to use xCAT and partimage to deploy images, and I don't see us moving completely away from bare metal images for many years. I have been told that more recent versions of xCAT 2 do have partimage support, but that you have to download that part separately. I have not looked in to this, but if it would be easy to switch VCL over to using that, it's possible we may do so in the future. Josh On Tuesday March 22, 2011, John Ma wrote: > My concerns to the pre-build images are > 1. Security - It's a black box. Will need our security officer's > approval for production use. > 2. Support - Will it still be available when we need to expand to other > hardware platform or next release of xCAT/partimage? > > John Ma > Marist College > > > > > From: Aaron Peeler > To: vcl-user@incubator.apache.org > Date: 03/22/2011 10:32 AM > Subject:Re: VCL2.2 + xCAT2.5 on bladecenter > > > > We like partimage-ng. > > But, we still use both partimage and partimage-ng. Our case is a little > unique, we started using xCAT1.3 with partimage, then added local > support ourselves because xcat 2.0 did not include imaging. > > Can you expand on what you feel uncomfortable about? Is it building them > or modifying, you should be able to mount them and view the contents. > > Aaron > > On 3/21/11 1:53 PM, John Ma wrote: > > Hi Aaron, > > > > I couldn't even finish configuring xCAT2.5 to capture image. What > > imaging module is in your setup, partimage, partimage-ng, systemimager > > or other? I have this resource > > https://cwiki.apache.org/VCL/adding-support-for-partimage-and-partimage-ng- > to-xcat-2x-unofficial.html , > > > but feel uncomfortable with the three pre-build stateless images. > > > > Thanks, > > John Ma > > Marist College > > > > > > > > > > From: Aaron Peeler > > To: vcl-user@incubator.apache.org > > Date: 03/21/2011 12:33 PM > > Subject: Re: VCL2.2 + xCAT2.5 on bladecenter > > > > > > > > > > Hello John, > > > > Yes. We're using xcat 2.5.1 and older versions
Re: VCL2.2 + xCAT2.5 on bladecenter
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 John, The stateless parts are not actually black boxes. They are gziped cpio files and are typical linux initrd and root filesystem images. You can do the following to extract them: mkdir /tmp/stateless cd /tmp/stateless mkdir initrd mkdir rootimg wget http://people.apache.org/~jfthomps/partimageng_xCAT2x/initrd.img wget http://people.apache.org/~jfthomps/partimageng_xCAT2x/rootimg.gz cd initrd zcat ../initrd.img | cpio --extract cd ../rootimg zcat ../rootimg.gz | cpio --extract # if you do this last command as a non- root user, you will get one error about not being able to create dev/null which is okay to ignore unless you want to modify and recreate the image You would actually need to go through this process if you have hardware that is not supported in these images. I only had the hardware at NCSU to develop on; so, they don't have a very big range of hardware support. If you decide to go ahead with using this stuff for partimage and need support for different hardware, I can guide you through adding it in to initrd and rootimg. If you wanted to modify and recreate the images after doing the above, you would do the following: cd /tmp/stateless/initrd find . | cpio -oc | gzip -c -9 > ../initrd-new.img cd /tmp/stateless/rootimg find . | cpio -oc | gzip -c -9 > ../rootimg-new.gz The two images really don't do much. Most of the work is done by the partimageng postscript. They really just provide a small install of linux on which the other stuff can run. As long as NCSU is using bare metal images, we'll still need a way to use xCAT and partimage to deploy images, and I don't see us moving completely away from bare metal images for many years. I have been told that more recent versions of xCAT 2 do have partimage support, but that you have to download that part separately. I have not looked in to this, but if it would be easy to switch VCL over to using that, it's possible we may do so in the future. Josh On Tuesday March 22, 2011, John Ma wrote: > My concerns to the pre-build images are > 1. Security - It's a black box. Will need our security officer's > approval for production use. > 2. Support - Will it still be available when we need to expand to other > hardware platform or next release of xCAT/partimage? > > John Ma > Marist College > > > > > From: Aaron Peeler > To: vcl-user@incubator.apache.org > Date: 03/22/2011 10:32 AM > Subject:Re: VCL2.2 + xCAT2.5 on bladecenter > > > > We like partimage-ng. > > But, we still use both partimage and partimage-ng. Our case is a little > unique, we started using xCAT1.3 with partimage, then added local > support ourselves because xcat 2.0 did not include imaging. > > Can you expand on what you feel uncomfortable about? Is it building them > or modifying, you should be able to mount them and view the contents. > > Aaron > > On 3/21/11 1:53 PM, John Ma wrote: > > Hi Aaron, > > > > I couldn't even finish configuring xCAT2.5 to capture image. What > > imaging module is in your setup, partimage, partimage-ng, systemimager > > or other? I have this resource > > https://cwiki.apache.org/VCL/adding-support-for-partimage-and-partimage-ng- > to-xcat-2x-unofficial.html , > > > but feel uncomfortable with the three pre-build stateless images. > > > > Thanks, > > John Ma > > Marist College > > > > > > > > > > From: Aaron Peeler > > To: vcl-user@incubator.apache.org > > Date: 03/21/2011 12:33 PM > > Subject: Re: VCL2.2 + xCAT2.5 on bladecenter > > > > > > > > > > Hello John, > > > > Yes. We're using xcat 2.5.1 and older versions > > > > Can you send your vcld.log file for the requestid it failed on? > > > > Aaron > > > > On 3/21/11 12:25 PM, John Ma wrote: > > > Hi community, > > > > > > I attempted and failed to setup xCAT 2.5 to work with VCL2.2 for bare > > > metal provisioning all within a bladecenter. Does anyone have a > > success > > > > story on this kind of setup yet? > > > > > > Thanks, > > > John Ma > > > Marist College > > > > -- > > Aaron Peeler > > Program Manager > > Virtual Computing Lab > > NC State University - -- - --- Josh Thompson VCL Developer North Carolina State University my GPG/PGP key can be found at pgp.mit.edu -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.16 (GNU/Linux) iEYEARECAAYFAk2IyvUACgkQV/LQcNdtPQMBgwCaApLT9Tla8NEwBDZNbJXSE3DN usYAn388FMgQRiSCmFqei9cCEKWIAfqk =Tv6U -END PGP SIGNATURE-
Re: VCL2.2 + xCAT2.5 on bladecenter
My concerns to the pre-build images are 1. Security - It's a black box. Will need our security officer's approval for production use. 2. Support - Will it still be available when we need to expand to other hardware platform or next release of xCAT/partimage? John Ma Marist College From: Aaron Peeler To: vcl-user@incubator.apache.org Date: 03/22/2011 10:32 AM Subject: Re: VCL2.2 + xCAT2.5 on bladecenter We like partimage-ng. But, we still use both partimage and partimage-ng. Our case is a little unique, we started using xCAT1.3 with partimage, then added local support ourselves because xcat 2.0 did not include imaging. Can you expand on what you feel uncomfortable about? Is it building them or modifying, you should be able to mount them and view the contents. Aaron On 3/21/11 1:53 PM, John Ma wrote: > Hi Aaron, > > I couldn't even finish configuring xCAT2.5 to capture image. What > imaging module is in your setup, partimage, partimage-ng, systemimager > or other? I have this resource > https://cwiki.apache.org/VCL/adding-support-for-partimage-and-partimage-ng-to-xcat-2x-unofficial.html , > but feel uncomfortable with the three pre-build stateless images. > > Thanks, > John Ma > Marist College > > > > > From: Aaron Peeler > To: vcl-user@incubator.apache.org > Date: 03/21/2011 12:33 PM > Subject: Re: VCL2.2 + xCAT2.5 on bladecenter > > > > > Hello John, > > Yes. We're using xcat 2.5.1 and older versions > > Can you send your vcld.log file for the requestid it failed on? > > Aaron > > > On 3/21/11 12:25 PM, John Ma wrote: > > Hi community, > > > > I attempted and failed to setup xCAT 2.5 to work with VCL2.2 for bare > > metal provisioning all within a bladecenter. Does anyone have a success > > story on this kind of setup yet? > > > > Thanks, > > John Ma > > Marist College > > -- > Aaron Peeler > Program Manager > Virtual Computing Lab > NC State University > -- Aaron Peeler Program Manager Virtual Computing Lab NC State University
Re: VCL2.2 + xCAT2.5 on bladecenter
We like partimage-ng. But, we still use both partimage and partimage-ng. Our case is a little unique, we started using xCAT1.3 with partimage, then added local support ourselves because xcat 2.0 did not include imaging. Can you expand on what you feel uncomfortable about? Is it building them or modifying, you should be able to mount them and view the contents. Aaron On 3/21/11 1:53 PM, John Ma wrote: Hi Aaron, I couldn't even finish configuring xCAT2.5 to capture image. What imaging module is in your setup, partimage, partimage-ng, systemimager or other? I have this resource https://cwiki.apache.org/VCL/adding-support-for-partimage-and-partimage-ng-to-xcat-2x-unofficial.html, but feel uncomfortable with the three pre-build stateless images. Thanks, John Ma Marist College From: Aaron Peeler To: vcl-user@incubator.apache.org Date: 03/21/2011 12:33 PM Subject: Re: VCL2.2 + xCAT2.5 on bladecenter Hello John, Yes. We're using xcat 2.5.1 and older versions Can you send your vcld.log file for the requestid it failed on? Aaron On 3/21/11 12:25 PM, John Ma wrote: > Hi community, > > I attempted and failed to setup xCAT 2.5 to work with VCL2.2 for bare > metal provisioning all within a bladecenter. Does anyone have a success > story on this kind of setup yet? > > Thanks, > John Ma > Marist College -- Aaron Peeler Program Manager Virtual Computing Lab NC State University -- Aaron Peeler Program Manager Virtual Computing Lab NC State University
RE: VCL2.2 + xCAT2.5 on bladecenter
Hi All, I just wanted to add to this and see if I was in the same pool as others: I was not able to get xCAT+VCL2.2 to work on a bladcenter where the Management node was a Virtual Machine in the bladecenter. I was told by the VMware that IPMI devices are masked by ESXi, and hence not available for the Guest OS to do IPMI. I am not sure if this is universally true for others? Best Regards, Amit > -Original Message- > From: Aaron Peeler [mailto:aaron_pee...@ncsu.edu] > Sent: Monday, March 21, 2011 12:33 PM > To: vcl-user@incubator.apache.org > Subject: Re: VCL2.2 + xCAT2.5 on bladecenter > > Hello John, > > Yes. We're using xcat 2.5.1 and older versions > > Can you send your vcld.log file for the requestid it failed on? > > Aaron > > > On 3/21/11 12:25 PM, John Ma wrote: > > Hi community, > > > > I attempted and failed to setup xCAT 2.5 to work with VCL2.2 for bare > > metal provisioning all within a bladecenter. Does anyone have a > success > > story on this kind of setup yet? > > > > Thanks, > > John Ma > > Marist College > > -- > Aaron Peeler > Program Manager > Virtual Computing Lab > NC State University > > > -- > BEGIN-ANTISPAM-VOTING-LINKS > -- > > Teach CanIt if this mail (ID 458005711) is spam: > Spam: > https://www.spamtrap.odu.edu/b.php?i=458005711&m=3adfd0a9415c&t=2011032 > 1&c=s > Not spam: > https://www.spamtrap.odu.edu/b.php?i=458005711&m=3adfd0a9415c&t=2011032 > 1&c=n > Forget vote: > https://www.spamtrap.odu.edu/b.php?i=458005711&m=3adfd0a9415c&t=2011032 > 1&c=f > -- > END-ANTISPAM-VOTING-LINKS
Re: VCL2.2 + xCAT2.5 on bladecenter
Hi Aaron, I couldn't even finish configuring xCAT2.5 to capture image. What imaging module is in your setup, partimage, partimage-ng, systemimager or other? I have this resource https://cwiki.apache.org/VCL/adding-support-for-partimage-and-partimage-ng-to-xcat-2x-unofficial.html , but feel uncomfortable with the three pre-build stateless images. Thanks, John Ma Marist College From: Aaron Peeler To: vcl-user@incubator.apache.org Date: 03/21/2011 12:33 PM Subject: Re: VCL2.2 + xCAT2.5 on bladecenter Hello John, Yes. We're using xcat 2.5.1 and older versions Can you send your vcld.log file for the requestid it failed on? Aaron On 3/21/11 12:25 PM, John Ma wrote: > Hi community, > > I attempted and failed to setup xCAT 2.5 to work with VCL2.2 for bare > metal provisioning all within a bladecenter. Does anyone have a success > story on this kind of setup yet? > > Thanks, > John Ma > Marist College -- Aaron Peeler Program Manager Virtual Computing Lab NC State University
Re: VCL2.2 + xCAT2.5 on bladecenter
Hello John, Yes. We're using xcat 2.5.1 and older versions Can you send your vcld.log file for the requestid it failed on? Aaron On 3/21/11 12:25 PM, John Ma wrote: Hi community, I attempted and failed to setup xCAT 2.5 to work with VCL2.2 for bare metal provisioning all within a bladecenter. Does anyone have a success story on this kind of setup yet? Thanks, John Ma Marist College -- Aaron Peeler Program Manager Virtual Computing Lab NC State University
VCL2.2 + xCAT2.5 on bladecenter
Hi community, I attempted and failed to setup xCAT 2.5 to work with VCL2.2 for bare metal provisioning all within a bladecenter. Does anyone have a success story on this kind of setup yet? Thanks, John Ma Marist College