Collect Stops / Init

[Bill Skulley_1]

I am running collect on my servers:
/usr/sbin/collect -i60,120 -f /var/adm/collect.dated/collect -H d0:5,1w -W 15m -M 10,15 -o n. On some of my servers, it will run forever without problems, on others, it will die for no apparent reason. I thought maybe I could set it up so init would respawn it if it died with an inittab entry collect:3:respawn:/usr/sbin/collect -i60,120 -f /var/adm/collect.dated/collect -H d0:5,1w -W 15m -M 10,15 -o n However, what this actually does is start 20 instances of collect and does *not* restart them when they are killed. What am I missing? Thanks.

[Abdul Rahiman]

I've had similar issues with collect on our GS160 running Tru64 5.1. It used to run fine on 4.x systems and other 5.x systems, but used to brake on just the GS160.
Did you observe the collect process taking up 100% cpu utilization before it died? In my case, collect process used to gradually suck a lot of CPU cycles before it died. Installing patch kit 5 resolved the issue. If you are not running PK5, take a look at the release notes regarding Patch# 823.0. Installing PK5 might help.

regds,
Abdul.

[Bill Skulley_1]

Thanks for your answer. We were seeing the 100% CPU use by collect a couple of years ago, but as you say it was fixed in one of the Patch Kits; we no longer see collect go "out of control", now it just disappears on some of our systems.