Subj : CPU Hog
To   : deon
From : Digital Man
Date : Tue Aug 09 2022 12:09 pm

  Re: CPU Hog
  By: deon to Digital Man on Tue Aug 09 2022 03:04 pm

 >  >  > Aug  5 00:35:54 d-11-1 synchronet: evnt BINKPOLL Running native timed
 >  >  > event: BINKPOLL
 >  >  > Aug  5 14:36:58 d-11-1 synchronet: evnt BINKPOLL Running native timed
 >  >  > event: BINKPOLL
 >
 >  >  > So if the first question is yes, then that would imply that binkit -p
 >  >  > does eventually exit.
 >
 >  > Yes.
 >
 >  > And for ever foreground "Running" log entry there should be a
 >  > corresponding log entry like this:
 >
 >  > BINKPOLL Timed event: BINKPOLL returned 0
 >
 > Ahh, OK - so it ran for 2hrs and 1 min the first time:
 >
 > Aug  5 02:36:58 d-11-1 synchronet: evnt BINKPOLL Timed event: BINKPOLL
 > returned 0
 > Aug  5 14:37:26 d-11-1 synchronet: evnt BINKPOLL Timed event: BINKPOLL
 > returned 0
 >
 > And the 2nd time it finished after 30s - because when it did poll my fido
 > uplink, it was stuck already (the uplink polled me at 14:33 - and thus my
 > connect to them got a "busy" message):

Okay, cool. But now you see how to determine how long an event took to run.

 > Aug  5 14:37:19 d-11-1 synchronet: evnt BINKPOLL Attempting poll for node
 > 3:633/280@fidonet
 > Aug  5 14:37:19 d-11-1 synchronet: evnt BINKPOLL JSBinkP/4 callout to
 > 3:633/280@fidonet started
 > Aug  5 14:37:20 d-11-1 synchronet: evnt BINKPOLL Connecting to
 > 3:633/280@fidonet at ftn633.vk3heg.net:24554
 > Aug  5 14:37:20 d-11-1 synchronet: evnt BINKPOLL Peer version:
 > binkd/1.1a-115/LinuxAug  5 14:37:20 d-11-1 synchronet: evnt BINKPOLL BinkP
 > got non-fatal error 'Secure AKA 3:633/509@fidonet busy' from remote:
 > 3:633/280@fido net,3:633/0@fidonet,21:1/195@fsxnet,39:901/280@amiganet,39:90
 > 1/0@amiganet,46:3/ 101@agoranet
 >
 > It appeared my fido polled me roughly every hour or so, and what I did
 > notice is the absense of a corresponding:
 >
 > BINKP service thread terminated (4 clients remain, 4 total, 60 served)
 >
 > for his polls, when his poll would have ended.
 >
 > Checking through my logs I see often "[3-6] clients remain" for the BINKP
 > service thread - so that indicates binkp sessions that havent yet exited?

Correct. You can also set a limit for incoming connections ("MaxClients") for each service, if you wished.

 >  > Well if you figure out how to reproduce the problem, let me know.
 >
 > I might be able to, as I think it is MTU related - and I can revert my MTU
 > configuration back to 1500 to try and get it to happen again. I'm not too
 > worried about fixing it, as it's no longer an issue over IPv4 - so I copied
 > the above for you in case there is another underlying issue with the BINKP
 > service thread and not exiting that you are interested in.

If it requires IPv6 or a custom MTU to reproduce, I'm unlikely to pursue the problem. Deuce might, I suppose, but he's been AWOL here for quite a while. Hopefully you can find a suitable solution with what you've learned.
-- 
                                            digital man (rob)

This Is Spinal Tap quote #4:
David St. Hubbins: He died in a bizarre gardening accident...
Norco, CA WX: 91.9øF, 29.0% humidity, 9 mph SSE wind, 0.00 inches rain/24hrs

.