Sorry, I forgot to put ML when replying I got further on this, for anybody interested:
There is a service (xcatpostinit1.service) that calls /opt/xcat/xcatdsklspost on computing node, this calls wget to fetch all the postscripts, if anybody has problems you can `sh -x /opt/xcat/xcatdsklspost` and see what is wrong. You can filter the wget command for `response` and make sure is everything downloading okay In my case the problem was simple simple simple, I got a postscript with wrong permissions so that wget get a 403 response, fixing permissions make everything work okay. I didn't try xcatprobe, I don't really know if it checks if postscripts are readable, if not, it would be nice to do it, IMHO BTW, It seems to work fine with symlinks ... Thanks Mark for pointing a direction, I wouldn't get it without your help Regards, Daniel ________________________________ De: Daniel Hilst <[email protected]> Enviado: terça-feira, 12 de novembro de 2019 20:01 Para: Mark Gurevich <[email protected]> Assunto: RE: [xcat-user] Troubleshooting postscripts? There is no /etc/rsyslog.d/remote.conf at all, But I got something interesting at compute node syslog, node001 xcat: ready node001 xcat: done node001 xcat.deployment: Call /opt/xcat/xcatdsklspost for stateless mode node001 xcat.deployment: trying to download postscripts... node001 chronyd[809]: Selected source 172.26.255.254 node001 chronyd[809]: System clock wrong by 1.350600 seconds, adjustment started node001 systemd: Time has been changed node001 chronyd[809]: System clock was stepped by 1.350600 seconds node001 xcat.deployment: failed to download the postscripts from the xCAT server for node node001.cluster.example.com node001 systemd: xcatpostinit1.service: main process exited, code=exited, status=1/FAILURE node001 systemd: Failed to start xcat service on compute node, the framework to run postbootscript and update node statu node001 systemd: Unit xcatpostinit1.service entered failed state. node001 systemd: xcatpostinit1.service failed. For some reason the download failed, but I can see the postscripts at /xcatpost, I'm using a symlink from /install/postscripts/versatushpc to another folder, and setting image as `chdef -t osimage -o ... -p postscripts=versatushpc/foo` this shouldn't be a problem, right? , ________________________________ De: Mark Gurevich <[email protected]> Enviado: terça-feira, 12 de novembro de 2019 19:26 Para: Daniel Hilst <[email protected]> Assunto: RE: [xcat-user] Troubleshooting postscripts? Is there a file "/etc/rsyslog.d/remote.conf" on the compute node with an entry to forward logs to management node ? Mark Gurevich Poughkeepsie Development Lab HPC Software Development - xCAT "If we knew what it was we were doing, it would not be called research, would it?" --Albert Einstein [Inactive hide details for Daniel Hilst ---11/12/2019 04:33:21 PM---Hi Mark, Executing it in the node works like a charm. I'm tr]Daniel Hilst ---11/12/2019 04:33:21 PM---Hi Mark, Executing it in the node works like a charm. I'm trying to understand the gears that are in From: Daniel Hilst <[email protected]> To: Mark Gurevich <[email protected]>, xCAT Users Mailing list <[email protected]> Date: 11/12/2019 04:33 PM Subject: [EXTERNAL] RE: [xcat-user] Troubleshooting postscripts? ________________________________ Hi Mark, Executing it in the node works like a charm. I'm trying to understand the gears that are in charge to run it at boot time. I just noted that `logger -t xCAT -p local4.info fooo` does not show at headnode /var/log/xcat/compute.log, even running on command line ________________________________ De: Mark Gurevich <[email protected]> Enviado: terça-feira, 12 de novembro de 2019 18:18 Para: xCAT Users Mailing list <[email protected]> Assunto: Re: [xcat-user] Troubleshooting postscripts? Try running the postscript yourself after the node is booted (https://xcat-docs.readthedocs.io/en/stable/guides/admin-guides/references/man1/updatenode.1.html#to-run-postscripts): * Make sure your postscript is in "/install/postscipts" on the management node and is executable * Invoke it with "updatenode <nodename> -P <postscript>" Mark Gurevich Poughkeepsie Development Lab HPC Software Development - xCAT "If we knew what it was we were doing, it would not be called research, would it?" --Albert Einstein [Inactive hide details for Daniel Hilst via xCAT-user ---11/12/2019 04:10:01 PM---Hi everybody, I'm trying to understand why my]Daniel Hilst via xCAT-user ---11/12/2019 04:10:01 PM---Hi everybody, I'm trying to understand why my postscript is not running. I set up a simple postscrip From: Daniel Hilst via xCAT-user <[email protected]> To: xCAT Users Mailing list <[email protected]> Cc: Daniel Hilst <[email protected]> Date: 11/12/2019 04:10 PM Subject: [EXTERNAL] [xcat-user] Troubleshooting postscripts? ________________________________ Hi everybody, I'm trying to understand why my postscript is not running. I set up a simple postscript to easy reproduction, my image object is like this https://gist.github.com/dhilst/2dc46ab433f17c07cce2c460fff05153 My postscript is in the same gist When booting the node, I follow all logs as `tail -F /var/log/xcat/*`. I can't see any information in the logs regarding postscript, I put the logs in the same gist too. When I log in the node I can see my script at `/xcatpost/versatushpc/` but it doesn't created the file /tmp/bootdate and no messages at computes.log so it seems to not being run. I'm using xCAT-2.15 recently released and CentOS 7 I don't really understand how and when postscripts run. They are started by a service? And syncfiles run before or after postscripts? I have a postscript that depends on syncfiles running before it, what would be the right approach? Use syncfiles EXECUTE directive? Regards, Daniel_______________________________________________ xCAT-user mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/xcat-user
_______________________________________________ xCAT-user mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/xcat-user
