[tor-bugs] #24158 [Core Tor/Tor]: I get this error "Looks like our kernel doesn't have the support for KIST anymore." on my relay

[Tor Bug Tracker & Wiki]

#24158: I get this error "Looks like our kernel doesn't have the support for KIST
anymore." on my relay
---------------------------+------------------------------------
 Reporter:  Dbryrtfbcbhgf  |          Owner:  pastly
     Type:  defect         |         Status:  assigned
 Priority:  Medium         |      Milestone:  Tor: 0.3.2.x-final
Component:  Core Tor/Tor   |        Version:
 Severity:  Normal         |     Resolution:
 Keywords:  tor-sched      |  Actual Points:
Parent ID:                 |         Points:
 Reviewer:                 |        Sponsor:
---------------------------+------------------------------------

Comment (by dgoulet):

 Replying to [comment:11 pastly]:
 > dgoulet: can you clarify what message(s) you want to keep and which you
 want to get rid of?
 >
 > We have
 >
 > `Scheduler type %s has been enabled` every time the scheduler switches
 cleanly. I think we should keep this as notice.

 Yes we keep this one.

 >
 > `Looks like our kernel doesn't have the support for KIST` every time we
 try to use KIST but have to fallback to KISTLite. Should happen at most
 once unless you play games with your torrc and HUPing. I think we should
 keep this as notice (maybe even warn, but we don't want it to be too scary
 since it may not indicate a problem).

 Yes keep this one.

 >
 > `Scheduler type KIST has been disabled by the consensus or no kernel
 support` which gets called every time we call `select_scheduler()`, have
 KIST in `Schedulers`, but can't use it. I think this should be moved to
 info or removed.

 This one I propose we keep it notice but print _once_ and after that we do
 not. However, in the case of no kernel support, this is detected at
 runtime and we do log notice about it when detected so we don't need it.
 Thus the only way to transition out of KIST while having KIST enabled in
 Schedulers then is through the consensus. We should log notice that it
 happened but only once, not everytime we get a consensus. And if we ever
 transition back to KIST after a while, we'll get the "Scheduler type %s
 has been enabled" so we'll be back on track.

 >
 > I will gladly take care of this.

 Makes sense to you?

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/24158#comment:12>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online