binkit timeouts/hanging #3
Description
I am still having (or started having again) issues with binkit hanging during
outbound sessions. I have a crontab script event that calls binkit every few
hours by running this single command:
touch /sbbs/data/binkit.now
That has been working without issue for months. Just recently, I have noticed
that binkit has been hung on the system. I notice it because QWK mail sessions
(users and nodes) come in, sit about 5 minutes, and then timeout without
creating a QWK OR they upload a REP which does not unpack.
If I 'Q'uit synchronet and start it up again, I will see a BINKIT error flash by
when it is going down, and will see the REP(s) uppack when I am bringing it back
up.
This is under linux, BTW.
Here is some log output. Note that the BINKIT session in question was started
at 1600 local time, so it sat that long until I hit 'Q' and it spat out these
messages (there was no error when everything stopped at 1600):
11/2 18:51:15 srvc 0040 BinkIT closing socket 0.0.0.0 port 24554
11/2 18:51:15 srvc 0041 BinkIT closing socket :: port 24554
11/2 18:51:15 srvc #### Services thread terminated (222 clients served)
11/2 18:51:15 evnt BINKIT Failed to receive the whole file
'/sbbs/temp/event/0000ffb5.sa0'.
11/2 18:51:15 evnt BINKIT Sending M_ERR command args: Forced Shutdown
11/2 18:51:15 term Waiting for events thread to terminate...
11/2 18:51:15 evnt BINKIT Attempting poll for node 314:314/0@pinet
11/2 18:51:15 evnt BINKIT JSBinkP/1.111 callout to 314:314/0@pinet started
11/2 18:51:15 evnt BINKIT Connecting to necrobbs.strangled.net:24556
11/2 18:51:16 evnt BINKIT Sending M_ERR command args: MD5 Required
11/2 18:51:16 evnt BINKIT !JavaScript /sbbs/exec/binkit.js line 255: TypeError:
bp.remote_addrs is undefined
11/2 18:51:16 evnt Timed event: BINKIT returned 0
Note that these messages only came out nearly 3 hours later, after I 'Q'uit synchronet and some of them could have been caused because I quit.
That 1600 session quit reporting any progress about 1602 with no error messages. Shortly after that, a QWK network "call" came in, which was apparently successful in generating a QWK
packet. The QWK network calls that were unsuccessful came later.
Shortly after I restarted synchronet, the next scheduled binkit session ran at
1900. It also stopped reporting any progress at about 1903.
For what it is worth, when bikit used to run without issue, it would usually be
done in under 2 minutes. It does not appear to be consistent as to which
session (i.e. node) it hangs during, although by law of averages it is usually with a mystic
system.
I checked my sbbsecho.ini and there do not appear to be any rogue entries at the
bottom as there were the last time I had any issues with binkit connections.
Version info, from the binkit.js file:
//