Semaphore management

rpetersonesq · Post by **rpetersonesq** » 28 Jan 2010 8:19

Receive an 'Internal Error' when trying to discover a new agent. The log entry states 'Failed to create semaphore with key=xxxxxxxx errno=28 No space left on device'. After running clen_sem I'm able to discover the agent. The issue is having to stop the server to run the clean utility. Is anyone aware of a way to configure semaphore usage so a periodic outage isn't necessary? The times between running the clean seems to be shrinking.

rpetersonesq · Post by **rpetersonesq** » 01 Feb 2010 9:23

Thank you for the response. I'm running EM 6.3.01.500 on Solaris 10, Oracle 10.2. I opened a ticket with BMC but they didn't really help. Had me run clean_sem which allowed me to discover an agent but I'm looking for a more permanent fix that doesn't require shutting down the server.

wrclark · Post by **wrclark** » 04 Feb 2010 12:02

I have the exact same issue, also on 6.3.1, Solaris 10, Oracle, etc and an open case with BMC on the topic. No firm resolution yet.

We have noticed that the number of semaphores used by the ctlm role seems to grow by 1 per day. The semaphore appears to be created by the CA process. clean_semaphores will resolve this.

There is a default limit of 120 semaphores per user [ or "project" to use a Solaris term] so you will start receiving error messages at that point. This limit can be increased by the system administrator if you are looking for a short term work-around. Doesn't help you much I know ...

Perhaps others have had a similar experience ?

philmalmaison · Post by **philmalmaison** » 04 Feb 2010 2:56

did you pass the initial checkreq ?
where you ok with preinstall oracle sacnning this kind of problem beetween installation?

isn't it an ulimit problem on your controlm user

anyway you have to adjust ulimit parameters and /etc/system if solaris

regards
philmalmaison