Hello Calvin,

not sure if I can fix your issue, but a few thoughts...

I would do an "lsdef nodename" and confirm what postscripts are set to run on 
the node to compared with what is mentioned in the xCAT postscripts table 
(tabdump postscripts).
As you have defined postscripts at a group level it will be useful to check 
that what node is actually inheriting from the postscripts table.
An lsdef against the node will validate what postscripts and in which order 
xCAT thinks it is going to run for that particular  node.

Some osimages are configured with  postscripts against the osimage (Lenovo do 
this for their DSS-G deployments)  so iwould check you image defintion:
Run "lsdef -t osiamge" to see you images and "lsdef -t osiuamge imagfename" to 
see the details.

Maybe you need to re-run the nodeset command (nodeset nodename 
osimage=imagename) when the node is deployed ? 
I think this makes no difference to postscripts but I cannot be 100% sure.

Also there may be some debug in /var/log/xcat/*log on your xCAT master node.
If you cannot see it try a tail -f on all those logs and redeploy the node 
again it may help you debug it further.

Regarding disableconsistentNICrename this is part of the xCAT code and not PCM 
as  I can set it on github:
https://github.com/xcat2/xcat-core/blob/master/xCAT/postscripts/disableconsistentNICrename

If this script is being called and it is not in xCAT maybe something has 
injected a startup service into systemd on the node or sometihng like 
/etc/rc.local I have not come across this before.
If you have dumped the entire xCAT database ("dumpxCATdb -p dirname") and it is 
not showing up anywhere then it cannot be xCAT running that script.
I do not recall that script being used by default in the past.

Regards,

David


-----Original Message-----
From: Calvin Dodge <[email protected]> 
Sent: Tuesday, December 17, 2019 9:40 PM
To: [email protected]
Subject: [External] [xcat-user] postscripts not being executed in order 
specified

We fixed this issues of nodes not booting completely, but now they don't set 
the deployment NIC's IP address.

The postscripts are defined only via groups, and I included "confignics" in the 
postscripts list.  But when a node is deployed, and I login via the console, 
the deployment NIC has no IP address assigned to it.

Perusing /var/log/xcat/xcat.log on the node in question shows the same order of 
postscripts as there were before I changed the postscripts assigned via the 
group in question, including at least one postscript which is not mentioned 
anywhere in the xCAT database or in any other script 
(disableconsistentNICrename).

Could this be something the Platform Cluster Manager (community edition, 
version 4.3.0) is doing behind the scenes? What else could be hijacking the 
postscript list?

Here's the current list of postscripts and postbootscripts.  The image being 
deployed is RHEL 7.6 stateless:

postbootscripts=setuppcm,syncfiles,ospkgs,otherpkgs,mountnfs,setupscratch,lsf_startup
postscripts=syslog,remoteshell,syncfiles,confignics,setminiuidgid,mkresolvconf,setupntp,setnetboot


Sincerely,

Calvin Dodge


_______________________________________________
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

Reply via email to