Why are both nodes on our production RAC swapping, yet they have loads of free memory?
There are several ways to configure the CPUs and memory in Alpha servers and I have no concept of the impact of these varying configurations. In my own simplistic view:
loads of free memory = no need to swap
Best get on to the guys in the know. See if they can give me a simple answer to this. It would be nice to get my hands on some of this free memory, but not if it’s going to cause a swap-frenzy 🙂
Cheers
Tim…
mumble mumble Working Set is too small mumble mumble. http://www.ttinet.com/tti/DLB_TUNE.HTML
It might help if you posted the OS you are using. Even some unix require you to tell how much memory to dedicate to the OS vs users.
Hi.
I wasn’t expecting help, it was just an expression of my frustration, hence the lack of detail. 🙂
This is a problem HP have been aware of for over two years and so far have done zilch.
If you are interested you might want to read this post:
http://www.uni-magdeburg.de/urzs/marvel/vmbug3.html
This guy has the same problem but much worse.
I guess our newly reconfigured hardware came with a few extra surprises. Unfortunately it’s one of those cases where it worked OK in development and test, but screws up in production.
Cheers
Tim…
Bye the way, it was Tru64 🙂
Cheers
Tim…