01:10 |
asciilifeform |
hmm shinohai let's try that again |
01:10 |
* |
asciilifeform foo |
01:10 |
* |
asciilifeform bar |
01:10 |
asciilifeform |
shinohai: same effect ? |
01:13 |
shinohai |
now I'm seeing you as hearsay again :/ |
01:15 |
shinohai |
http://btc.info.gf/uploads/PEST2.png |
| |
↖ |
01:30 |
asciilifeform |
hmm likely bug |
01:30 |
asciilifeform |
awt ^ |
01:31 |
asciilifeform |
and wat's with the join/part thing.. |
01:31 |
* |
asciilifeform not saw any corresponding sequence of join/part for shinohai |
01:32 |
shinohai |
My bot was doing same thing before I shut it down, dunno why the join/part thing kept happening. |
| |
~ 12 hours 55 minutes ~ |
14:28 |
awt |
asciilifeform, shinohai: http://logs.bitdash.io/log-search?q=peer_offline_interval_seconds&chan=pest |
14:28 |
awt |
asciilifeform: what evidence is there that this is a bug? |
14:28 |
asciilifeform |
awt: hmm what's that knob set to by default ? |
14:29 |
shinohai |
ack awt, I had forgotten to set that. |
14:29 |
shinohai |
Still seeing asciilifeform as hearsay tho. |
14:29 |
asciilifeform |
seeing shinohai as peered tho |
14:30 |
shinohai |
heh that time you came through normally. |
14:30 |
awt |
asciilifeform: try %knob |
14:30 |
asciilifeform |
awt: 60 |
14:30 |
awt |
Yeah that's the default |
14:30 |
asciilifeform |
these join/parts defo not 60s apart |
14:30 |
bitbot |
Logged on 2022-04-19 01:15:59 shinohai[asciilifeform]: http://btc.info.gf/uploads/PEST2.png |
14:31 |
asciilifeform |
(and no such join/parts seen on asciilifeform's end fwiw) |
14:31 |
asciilifeform |
hence suspected bug |
14:32 |
shinohai |
imma try setting that knob to large # guess I can cobble together filter for weechat if dunwork. |
14:32 |
asciilifeform |
shinohai: notice they weren't 60s apart tho |
14:32 |
asciilifeform |
and no join/part noise b/w/ asciilifeform showing up on shinohai's end as peered vs hearsay |
14:32 |
shinohai |
Yeah the join/parts all occur on my end @ completely random times. |
14:32 |
asciilifeform |
so difficult to see how the knob pertains |
14:33 |
asciilifeform |
fwiw shinohai has been showing up on asciilifeform's console as peered ever since this point |
14:33 |
bitbot |
Logged on 2022-04-18 22:15:59 shinohai[asciilifeform]: aha! now works, you just "joined". |
14:33 |
awt |
asciilifeform: are you familiar with how the decision is made as to when to send a join/part? |
14:34 |
asciilifeform |
awt: iirc is when no packets recvd during peer_offline_interval_seconds neh |
14:34 |
asciilifeform |
(then , 'part') |
14:39 |
awt |
presence check is made every 5 seconds |
14:40 |
awt |
So join could come much sooner than 60s |
14:41 |
asciilifeform |
http://btcbase.org/patches/9979-presence#L165 hm |
14:41 |
asciilifeform |
join yes. but why the part ? |
14:42 |
awt |
a part indicates (duriing a presence check) that the at hasn't been updated for > 60s. |
14:44 |
asciilifeform |
awt: do i misread or is the timestamp in fact only updated when there's a logline rather than when any valid packet? |
14:47 |
asciilifeform |
or hm, superficially correct. but still suspect that update_at was not being called, for some reason, on shinohai's station |
14:48 |
awt |
asciilifeform: see station.py:141 |
14:48 |
awt |
should be called on IGNORE |
14:48 |
asciilifeform |
http://btcbase.org/patches/9978-bugfixes/tree/blatta/lib/station.py << lol 404 ( hey phf ?? ) |
| |
↖ |
14:48 |
* |
asciilifeform goes to look in local copy |
14:53 |
asciilifeform |
awt: loox like in get_at , the time is converted from utc to local, but in in handle_is_online, time.time() used ? |
14:56 |
* |
asciilifeform currently suspects that bug triggered on machine where local clock not set to utc. but not yet found concrete proof |
15:00 |
awt |
asciilifeform: possibly but pestbot is running on a machine with local time not set to utc and I don't see these issues. |
15:00 |
asciilifeform |
hmm |
15:14 |
* |
signpost just set his server to UTC to see if that caused peering with billymg to work. |
15:14 |
signpost |
doesn't look like it, current active peers are asciilifeform, awt, jonsykkel, mod6, shinohai. |
15:16 |
* |
shinohai has no other peer in list currently but asciilifeform however sees signpost fine. |
15:16 |
awt |
signpost: possible debugging technique for peering issues could be do backup blatta db, remove all other peerings, then send to the other party to debug locally |
15:21 |
signpost |
will give that a try when he pops on next |
| |
~ 1 hours 28 minutes ~ |
16:49 |
billymg |
!c uptime |
16:50 |
billymg |
meh, still nothing. tried opening crawlerbot's port via ec2's "security groups" UI |
16:50 |
billymg |
so not that |
16:58 |
billymg |
signpost: still a mystery to me why we're not peered but at this point i'm ok to wait for it to "fix itself" (via whatever random change in system or config finally shakes it loose) |