+orignal
+postman
+qend-irc2p
+sourceress
Arch
Birdy
Danny
Irc2PGuest1884
Irc2PGuest30010
Irc2PGuest36077
Irc2PGuest49364
Irc2PGuest51117
Irc2PGuest6564
Irc2PGuest65656
Irc2PGuest67278
Irc2PGuest83482
Irc2PGuest99117
MatrixBot
Onn4l7h
Over
R4SAS
RN_
Sleepy
T3s|4
Teeed
Yotsu
__bob_
aargh3
ac9f
acetone_
ahiru
anontor
b3t4f4c3__
cims_
dr4wd3_
duanin2
eche|off
f00b4r_
hababam
hagen_
leopold
makoto
marek
marek22k
n2
nilbog
noidea
not_bob_afk2
nyaa2pguy
o3d3_
poriori
profetikla
r00tobo
rapidash
solidx66
stormycloud[m]
test7363673
uop23ip
urist_
user_
zelgomer
zzz
zzz
eyedeekay, gitea down
eyedeekay
It was back by the time I looked at it
zzz
ok, the message was a little different than usual
zzz
remote:
zzz
remote: error:
zzz
remote: error: Internal Server Connection Error
zzz
remote: error:
zzz
error:
zzz
error: Failed to execute git command
zzz
error:
zzz
remote: . Processing 1 references
zzz
send-pack: unexpected disconnect while reading sideband packet
zzz
error: error in sideband demultiplexer
eyedeekay
Hm, that's new. All this stuff that's been seems to come down to a few locking issues in the interaction between git and gitea and instinctively, this seems like maybe related manifestation of that thing. I've been archiving the log every time it goes down to find it but most problems just go away after cleaning up the locks and restarting
eyedeekay
The locking is also demonstrably the cause, malfunctions begin to occur when git operations compete for a lock
eyedeekay
So now we're automatically cleaning locks every restart, and automatically restarting when lock competition causes a timeout, which takes about 2-5 minutes because it looks for false positives before killing/restarting and takes a backup, with restarts happening about 1-2 times a day, if you hit it in exactly that window which I think you did it probably acted weird
eyedeekay
See if I can find you in the logs...
zzz
no need, I leave it to you
eyedeekay
Well I'm curious, and if I can find the event maybe it's another clue