[Chugalug] What are the $%& boxes waiting for? :(
flushy at flushy.net
Tue Mar 25 14:17:17 UTC 2014
Do your disk stats correlate to your wait states?
CPU wait pretty much means waiting on IO. Specially, the process or thread is inside a kernel function that is in turn waiting on data. It's normally preempt-able, but in some cases it's not -- or it at least doesn't act like it: due to race conditions or dependent threads needing the data that the other thread is waiting on.
Additionally, your loadaverge should be high if your CPUs are in wait state a lot.
Sent from my iPhone
> On Mar 25, 2014, at 9:29 AM, kitepilot at kitepilot.com wrote:
> These are single boxes, doing different tasks (email, DNS, database) with no other relationship between them than they are located in the same building, connected to the same network and providing the services enumerated above (plus some other things) while running some Debian, some LFS, and several releases (6, 7, etc).
> They all exhibit the same behavior under nmon, LOTS of waiting state in the processors.
> I can't figure out what in the $#%&@ WORLD! the stupid processors are waiting for...
> I can't even come up with a meaningful Google search! :(
> Jonathan Calloway writes:
>> You got clustered or doing distributed jobs or something like that? Do you have some sort of agent or so ethi g else common on all of them? Jonathan Calloway Sent from my iPad
>>> On Mar 24, 2014, at 4:19 PM, flushy at flushy.net wrote: CPU IO wait normally means "waiting on IO". IO in this instance normally means "waiting on disk" It could mean "waiting on network" -- but normally network connections have time outs. --b _______________________________________________
>>> Chugalug mailing list
>>> Chugalug at chugalug.org
>> Chugalug mailing list
>> Chugalug at chugalug.org
> Chugalug mailing list
> Chugalug at chugalug.org
More information about the Chugalug