Bug 512616

Summary: squid wakes up the CPU when doing nothing
Product: [Fedora] Fedora Reporter: Edwin ten Brink <fedora>
Component: squidAssignee: Jiri Skala <jskala>
Status: CLOSED NEXTRELEASE QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: low Docs Contact:
Priority: low    
Version: 11CC: aglotov, drees76, henrik, jonathansteffan, jskala, mnagy
Target Milestone: ---Keywords: Upstream
Target Release: ---   
Hardware: All   
OS: Linux   
URL: http://www.squid-cache.org/bugs/show_bug.cgi?id=2011
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2009-07-22 03:17:24 EDT Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---
Bug Depends On:    
Bug Blocks: 204948    

Description Edwin ten Brink 2009-07-19 16:57:35 EDT
Description of problem:
Even without any connected clients, squid wakes up the CPU 100 times per second according to powertop. This does not seem necessary.

Version-Release number of selected component (if applicable):
$ rpm -q squid
Comment 1 David Rees 2009-07-20 14:00:26 EDT
Confirmed here, also affects F10 since it has the same version of squid.  Powertop reports this:

  12.3% (100.8)             squid : __mod_timer (process_timeout)

Stracing squid shows this going on 100 times a second:

epoll_wait(6, {}, 16384, 10)            = 0
Comment 2 Henrik Nordström 2009-07-20 15:56:58 EDT
Upstream are aware of this. Don't remember right now why it needed to do this other than that it's a workaround for a timing race and that it is not likely to be fixed in 3.0.

3.1 will behave better and will be available in F-12.
Comment 3 Edwin ten Brink 2009-07-21 15:38:10 EDT
(In reply to comment #2)
Found your original posting on squid-users mailing list in October 2008:

(In reply to comment #1)
David, you seem to be right about the polling:
Comment 4 Henrik Nordström 2009-07-21 16:02:26 EDT
The polling is a red herring (just an effect, not the cause), the issue is elsewhere, with the polling just plastering over it..

Linked to upstream bug report (closed).
Comment 5 Edwin ten Brink 2009-07-22 03:17:24 EDT
(In reply to comment #4)
> Linked to upstream bug report (closed).  

Ok, thanks. Let's close this one for now then with CLOSED - NEXTRELEASE (squid 3.1) as it appears to have been resolved indeed from the bug report at http://www.squid-cache.org/bugs/show_bug.cgi?id=2011