On Mon, 27 Mar 2017, Joshua Baker-LePain wrote:
Investigating my odd 'some commands don't generate output' issue revealed a
couple other issues. First, on my old cluster (running OGS 2011.11p2), only
a very few hosts are admin hosts -- notably, the exec nodes are not. On those
nodes, 'qconf -s' commands still work (i.e. you can list the configuation),
but 'qconf -m' commands don't, as you would expect. On my new cluster
running SoGE 8.1.9, even 'qconf -s' commands fail on non-admin hosts. If the
exec nodes are not admin hosts, this breaks the init.d script, which uses
'qconf -sconf' in the 'GetExecdSpoolDir' function. Was the change to disallow
'qconf -s' commands on non-admin hosts deliberate? I'm not a fan of making
every exec host an admin host -- it just seems unnecessary. Is there another
option?
Hi Joshua,
This one is down to a fix for this ticket:
https://arc.liv.ac.uk/trac/SGE/ticket/1547
Which was reverted after the release of 8.1.9:
https://arc.liv.ac.uk/trac/SGE/changeset/4912/sge
If you want this fixed ahead of the next release, I'm afraid you'll have
to apply the patch and rebuild.
Second, in looking through the sgemaster logs, I see a lot of lines like
these:
03/25/2017 00:15:27|event_|qb3-mgmt-1|E|no event client known with id 12 to
process acknowledgements
03/25/2017 01:17:00|worker|qb3-mgmt-1|E|no event client known with id 11 to
shutdown
I see a lot more of the first line than the second, and the IDs bounce around
in the range of 11-15. Google doesn't yield much -- any ideas?
...
Not seen that one and don't have time to look right now, sorry. You've not
still got some execd's from an old release kicking around, have you?
Mark
_______________________________________________
users mailing list
users@gridengine.org
https://gridengine.org/mailman/listinfo/users
