Re: Mysterious CPU eater in 3.0

 afaik all network ops should be accounted by profiler, by means of
 w/read_handlers in comm_poll at least. But I haven't looked after 3.0
 How large is prof_unaccounted during normal high load? If its more
 than 1-5% then something substantial is unaccounted for.

 Is it possible that more than 256 clients are fetching same object?
 any external acl or redirectors? (they might be unccounted. not checked)

 ps. you see cpu eater in 30sec avgs probably because you can't get
 reply from squid until it gets 'moment of rest' between racing, and
 then you see avgs of those rest moments.

On 3 Jun 2003 at 9:23, Joe Cooper wrote:

> One more data point:
>
> Reducing disk/memory usage has no impact--the problem still occurs,
> possibly less frequently. The machines have ample memory at 2GB and
> even with the 60GB cache_dir and 256MB cache_mem, there is 700+MB of
> available memory.
>
> The disk I/O type has no impact. aufs and ufs are both similarly effected.
>
> I kind of suspect server side network issues, as it seemingly happens
> more frequently when the byte hit ratio is extremely bad (due to
> range_offset_limit and quick_abort_min being set to -1).
> --
> Joe Cooper <joe@swelltech.com>
> Web caching appliances and support.
> http://www.swelltech.com
>
>
Received on Tue Jun 03 2003 - 11:30:47 MDT