OpenVZ Forum


Home » General » Support » numsiginfo resource growing
numsiginfo resource growing [message #32457] Sat, 09 August 2008 08:19 Go to next message
alevchuk is currently offline  alevchuk
Messages: 22
Registered: February 2007
Location: University of California,...
Junior Member
Dear Community,

I have a 64GiB RAM box running the 2.6.24-openvz-24-004.1d1-amd64 kernel.


My numsiginfo resource is always growing.

It grows linearly, gets to about 45, zeroes-out, and then continues growing again.

You can see the graphs here: http://biocluster.ucr.edu/munin/girke/owl.girke-vebc_numtcps ock_numflock_numpty_numsiginfo_numothersock_numiptent_105.ht ml


Does anyone experience the same behavior?

My load of the system is also growing, it is now 72 (!).


Is the high load a result of the siginfo?


The siginfo is the only resource that I can see abused on the system. Perhaps I'm not looking in the right place. Perhaps I'm just blind.


Alex

Report message to a moderator

Re: numsiginfo resource growing [message #32551 is a reply to message #32457] Fri, 15 August 2008 11:50 Go to previous messageGo to next message
maratrus is currently offline  maratrus
Messages: 1495
Registered: August 2007
Location: Moscow
Senior Member
Hi,

Quote:


My load of the system is also growing, it is now 72 (!).



What do you mean? How do you measure system load?

Report message to a moderator

Re: numsiginfo resource growing [message #32552 is a reply to message #32551] Fri, 15 August 2008 13:17 Go to previous messageGo to next message
alevchuk is currently offline  alevchuk
Messages: 22
Registered: February 2007
Location: University of California,...
Junior Member
`top` or `uptime` in Debain display the "load average".

I'm not 100% sure what is the formula, but usually on a healthy box this value does not got higher then 3.

Alex

Report message to a moderator

Re: numsiginfo resource growing [message #32554 is a reply to message #32552] Fri, 15 August 2008 14:01 Go to previous message
maratrus is currently offline  maratrus
Messages: 1495
Registered: August 2007
Location: Moscow
Senior Member
Hi,

load average includes the processes in a running state but also in a uninterruptible sleep state (in most cases it concerns with IO activity).
So we can check these two groups of processes (utility ps: D and R states).
Are there processes that are stuck inside kernel (ps: wchan).
vmstat - is also might be useful.

Report message to a moderator

Previous Topic: autofs in VE under Lenny
Next Topic: Kernel 2.6.26 status
Goto Forum:
  

[ Syndicate this forum (XML) ] [ RSS ] [ PDF ]

Current Time: Fri Nov 07 04:21:26 GMT 2025

Total time taken to generate the page: 0.10917 seconds
Powered by FUDforum Powered by Virtuozzo Containers