How often to reboot

Good question! You'll probably get much debate on this one.

I used to support 275 Sun systems networked together. I've since consolidated down to about 15 larger servers. Anyway, I found that rebooting a server is not always a good reason to fix stability issues. Instead, try to find out what specifically is unstable and restart that specific process or application. If you can, find out if there are any patches to fix any "stability" issues and apply those.

Past system administrators where I work were rebooting weekly (that was a chore!). This lead to the vendors and application administrators developing sloppy code. They were leaving memory leaks in their applications and, if their application was buggy, a reboot was a common fix.

Now I reboot only when necessary- kernel/security patches, hardware changes, HA tests, etc. I found that although some people disliked this approach at first, it lead to much higher system availability.

I hope you get plenty of responses to this question. There are a lot of views on this issue.

spamly at 2007-7-5 18:06:33 >

Thanks for the comments. I agree that the stability issues are generally with other applications vs. the kernal processes. I have always found Solaris to be very stable. Our issue is with these 3rd party apps we require yet we are still at the mercy of the vendor in some cases. I have heard recommendations similar to once a quarter obviously depending on the tolerance for downtime.

Thanks again

mscott0665 at 2007-7-5 18:06:33 >