Thanks for your advice, Nuno.
| nuno wrote on Fri, 15 January 2010 08:58 | | Quote: |
prometheus:
top - 08:06:00 up 3 days, 14:46, 2 users, load average: 0.00, 0.00, 0.00
Tasks: 140 total, 1 running, 137 sleeping, 1 stopped, 1 zombie
Cpu(s): 0.3%us, 0.2%sy, 0.0%ni, 99.5%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st
Mem: 3893072k total, 3811312k used, 81760k free, 427236k buffers
Swap: 1951856k total, 156k used, 1951700k free, 2782380k cached
|
This one is doing nothing at all.
|
True, I just migrated these machines from yorktown last night.
| Quote: |
and yorktown:
top - 08:06:17 up 14:07, 2 users, load average: 1.34, 1.08, 0.73
Tasks: 209 total, 1 running, 208 sleeping, 0 stopped, 0 zombie
Cpu(s): 24.8%us, 2.3%sy, 0.0%ni, 72.6%id, 0.2%wa, 0.0%hi, 0.2%si, 0.0%st
Mem: 2073688k total, 1933728k used, 139960k free, 103920k buffers
Swap: 7815580k total, 456k used, 7815124k free, 871652k cached
|
This one has a 0.75 15mins load and has half of the RAM free. So both can be loaded further.
If you have problems i sugest inspecting /proc/user_beancounters
Regards,
[/quote]
I have done that, however, I really havent gotten my brain around how you tweak things in a rational manner. I have taken to just tuning a parameter until it quits complaining. For instance, I have the following issues with the machines on yorktown:
Wiki server:
lockedpages 0 107 256 256 3
(For this, do I vzctl set <CONTAINER> --lockedpages 120:256 --save ?)
The same goes for some of the other parameters. The docs are somewhat vague when it comes to the optimum settings for them. For instance, I have three VMs that have failcnts of 3 for lockedpages, and one (the log server/splunk box) that has failcnts of 33 and 38 for privvmpages and tcpsndbuf respectively. privpages on this box is set for 123768 261816 262144 262144. tcpsndbuf is 232448 1722432 1720320 2703360.
is there a "for dummies" book on setting parameters based on user_beancounters?
Thanks,
--vr
|
OpenVZ Forum
Members
Search
Help
Register
Login
Home
