Hi Hurng, all,
Ribble is intended as a login/log-through system, and not intended for running analysis. Over the past months, ribble was frequently loaded by analysis jobs and completely unresponsive to anything else, so Ton and the Helpdesk put in some helpful limits to discourage such activity. The intention is to log through to your own workstation or a group server and do your analysis there. So ulimit -a now has
max user processes (-u) 16
and no more than 4 simultaneous logins on this host. At least now ribble is usable again as a login server ;-) This was in /etc/motd for the last few weeks, but you just missed it for a very good reason :-) The max-user processes of course triggers the fork(2) error.
The help desk should have more info.
Cheers, DavidG.
PS: you still have process running that is eating 100% CPU. I have seen this happen a lot of times, as apparently the Atlas CMT version has a bug that causes this process: sh -c /data/atlas/offline/14.2.20/CMT/v1r20p20080222/mgr/cmt_gcc_version.sh | /data/atlas/offline/14.2.20/CMT/v1r20p20080222/mgr/cmt_filter3_version.sh to hand and eat CPU cycles indefinitely. The helpdesk already killed like processes from many Atlas users :-) Any idea who CMT exhibits that behaviour?
Hurng-Chun LEE wrote:
Hello,
just recently, users on ribble get from time to time the error message like: "fork: Resource temporarily unavailable". I also get this error when ssh-ing to ribble.
This is quite abnormal. Could you please have a look? Thanks!
Cheers, Hurng _______________________________________________ Ct-grid mailing list Ct-grid@nikhef.nl https://mailman.nikhef.nl/cgi-bin/listinfo/ct-grid