[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]
Re: [tor-bugs] #25461 [Core Tor/Tor]: main event-loop spins consuming 100% of a CPU core at times
#25461: main event-loop spins consuming 100% of a CPU core at times
--------------------------+------------------------------------
Reporter: Dhalgren | Owner: (none)
Type: defect | Status: new
Priority: Medium | Milestone: Tor: 0.3.4.x-final
Component: Core Tor/Tor | Version: Tor: 0.3.3.3-alpha
Severity: Normal | Resolution:
Keywords: performance | Actual Points:
Parent ID: #25500 | Points:
Reviewer: | Sponsor:
--------------------------+------------------------------------
Comment (by Dhalgren):
A theory randomly popped into my mind as to the possible cause:
Use a Linux `tc` police rule to limit bandwidth on ingress and noticed the
CPU saturation problem shows up, at least at times, when average ingress
traffic is near the tc police limit but not in excess of it. Probably
packet discards from tc-police stalls many TCP connections in the middle
of app-layer buffer deliveries. Is it possible that libevent or a tor
dameon handler spins when incomplete buffers are pending assembly rather
than blocking?
Did not investigate code--is just an idea.
--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/25461#comment:5>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online
_______________________________________________
tor-bugs mailing list
tor-bugs@xxxxxxxxxxxxxxxxxxxx
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-bugs