00:01:16Chris5010 (Chris5010) joins
02:26:22Exorcism8 (exorcism) joins
02:28:01Exorcism quits [Read error: Connection reset by peer]
02:28:01Exorcism8 is now known as Exorcism
03:20:07Peroniko joins
03:22:15RetiredTurtle quits [Ping timeout: 265 seconds]
06:27:53project10 quits [Ping timeout: 252 seconds]
06:29:17project10 (project10) joins
06:36:08datechnoman quits [Quit: The Lounge - https://thelounge.chat]
06:38:20datechnoman (datechnoman) joins
06:38:29@flashfire42 quits [Client Quit]
06:38:29kiska quits [Client Quit]
06:40:09flashfire42 joins
06:41:24kiska (kiska) joins
07:00:45@ChanServ sets mode: +o flashfire42
11:44:14Exorcism quits [Remote host closed the connection]
11:45:24Exorcism (exorcism) joins
12:45:41Peronikola joins
12:48:29Peroniko quits [Ping timeout: 252 seconds]
15:04:32datechnoman quits [Ping timeout: 265 seconds]
15:19:46kiryu quits [Client Quit]
15:28:45datechnoman (datechnoman) joins
15:34:13Peronikola is now known as Peroniko
17:21:38Peronikola joins
17:24:02Peroniko quits [Ping timeout: 252 seconds]
17:29:30kiska quits [Client Quit]
17:29:30@flashfire42 quits [Client Quit]
17:32:19flashfire42 joins
17:33:33kiska (kiska) joins
19:03:27Exorcism quits [Remote host closed the connection]
19:04:11Exorcism (exorcism) joins
19:20:30Exorcism quits [Remote host closed the connection]
19:22:57Exorcism (exorcism) joins
20:34:05Exorcism quits [Remote host closed the connection]
20:34:58Exorcism (exorcism) joins
20:46:17Maturion joins
20:46:29Peronikola is now known as Peroniko
20:51:32Exorcism quits [Remote host closed the connection]
20:53:01Exorcism (exorcism) joins
21:12:38Maturion quits [Remote host closed the connection]
21:14:50<@arkiver>pokechu22: AB still gets banned on 2 second delay?
21:15:56<pokechu22>It's kinda random, but the job does get banned with the random delay from 1s to 2s sometimes. I *think* it happens when it retrieves a bunch of pages that don't exist in a row but I'm not 100% sure. Other jobs were able to run at a constant 1s delay without getting banned. (Though latency might factor into this as well which I'm not sure of)
21:16:42<@arkiver>could AB not be respecting the 2 second pause when a page throws some error?
21:17:24<@JAA>AB should always respect the delay setting.
21:17:38<pokechu22>I'm pretty sure it does, but there might be some odd behaviors with the way the redirects work or something. It definitely waits between each step of the redirect
21:17:53<imer>pokechu22: i can burn through some ips I can't use for archival anyways if we want to test things?
21:18:05<@JAA>But latency could play a role in this, too, if those 404s are returned much faster than 200s, for example.
21:18:30<imer>do we know how quick/long the bans are?
21:18:54<pokechu22>24 hours
21:22:32<imer>and a ban was a connection timeout?
21:23:17<imer>for test urls, should I just use the ones from above https://coturnix.monsite-orange.fr/ (working) and https://coturnix.monsite-orange.fr/son/historique.mid 404? or is "non existant pages" something else?
21:52:02<pokechu22>Yeah, connection timeouts. I think I also got connections refused on a local connection but I'm not 100% sure
21:52:16<pokechu22>The nonexistent pages thing is a guess that I'm not 100% sure of
21:56:11<pokechu22>The refused connections might have been with a delay shorter than 1s actually
21:56:37<pokechu22>I think that was 500ms or 500ms-1000ms
22:27:28qwertyasdfuiopghjkl quits [Remote host closed the connection]
22:29:45<imer>ban only seems to apply to new connections, doesnt kill the old ones
22:32:49<imer>you can burst a lot from the looks of it, so this might take a while
22:41:37<imer>Worked 0000429, 1.0 req/s, 433s elapsed and then it timed out
22:49:04<imer>Worked 0000480, 1.5 req/s, 327s elapsed this one started timing out at ~430 requests again but recovered temporarily and got completely blocked at ~480
22:52:20<imer>i'm guessing there's both a min. time and min. req number component to this, some of the earlier tests I went way faster (> 1k requests), but i'll retest that
22:52:43<imer>these are all on 404 urls for now
22:53:03<pokechu22>Are you measuring delay between requests, or between starting each request (disallowing concurrent requests)?
22:53:21<pokechu22>and are you following the redirects for 404 URLs?
22:53:43<pokechu22>for a DPoS job we'd probably not want to follow them (but AB will always follow them and there's no option to disable it)
22:54:13<imer>I've got (somewhat bad) code that adds an item to queue and then sleeps to match the rate, the x req/s is calculated in log output is calculated, so thats the real rate
22:54:17<imer>not following redirects
22:54:50<imer>i've got a number of workers, so if requests take longer than the rate it'll do multiple at once
22:55:13<imer>Worked 0000099, 0.5 req/s, 201s elapsed -> timeout
22:55:28<imer>gonna try the same on a non-404 now
22:55:41Peroniko quits [Ping timeout: 252 seconds]
22:55:50<imer>actually, going to make a proper build and run multiple at once
22:55:58<imer>(on different ips of course)
22:56:36<pokechu22>One other thing that'd be useful is determining rate-limits for stuff on woopic
22:56:49<pokechu22>I *think* they tie into the same rate limit but I'd need to double-check
23:00:14<pokechu22>hmm, actually, I don't have data either way on that one
23:00:57<pokechu22>it never ended up requesting anything on woopic while it was banned
23:01:28<pokechu22>but I guess *you* could easily test it by trying to load e.g. https://monsite.woopic.com/383/f/300x/p/bienetre-artisanat02/img/de96401a61f9936cc59ebf45fd037fd5.jpg on a banned IP
23:02:19Peroniko joins
23:13:02<imer>ban aplies to both
23:36:18<imer>I think 404's have a higher penalty for whatever they're using, non-404 ones are definitely going longer without ban so far
23:41:02<imer>yeah, got a ban on 404-only at 4s/req
23:49:29<imer>i did get one ban on 200 at 1req/s (after 800req), so probably have to go slightly slower, just not sure what the strategy for 404s should be, quite slow probably to make sure there's no bans
23:49:47<imer>and by 404 I mean the redirect
23:49:59<pokechu22>Does the penalty behavior change if you follow the redirects?
23:50:19<imer>that'd be quite advanced
23:50:38<imer>will give these a bit longer and then i'll try
23:51:13<pokechu22>Well, it'd be as simple as having a lighter penalty for the redirects themselves (e.g. a 1s penalty) and then things would average out to a lower penalty overall maybe