02:48:50 Monster database of master branch on crawl.develz.org updated to: 0.23-a0-650-ga4d50d4 03:31:18 Experimental (bcrawl) branch on crawl.kelbi.org updated to: 0.23-a0-1319-g59089ec972 04:16:56 -!- amalloy is now known as amalloy_ 07:54:21 -!- muffindrake1 is now known as muffindrake 09:24:32 Trove: unusable armour for Ogre 13https://crawl.develz.org/mantis/view.php?id=11816 by Nebukadnezar 13:18:14 -!- amalloy_ is now known as amalloy 14:37:11 MauveAvengr (L3 OpBe) Crash caused by signal #15: Terminated (D:1) 14:37:27 cerohero32 (L7 GrFi) Crash caused by signal #15: Terminated (D:5) 14:39:56 has anyone tried to figure out the cause of Sequell slowness recently? i'm setting up a local Sequell instance to see if i can tell what's taking so long, and if anyone has tried i'd be happy to know where to start or what to skip 17:55:24 Flugkiller (L25 GnAM) ASSERT(!invalid_monster(&mons)) in 'mon-death.cc' at line 1900 failed. (Vaults:5) 18:23:47 ASSERT(!invalid_monster(&mons)) in 'mon-death.cc' at line 1900 failed 13https://crawl.develz.org/mantis/view.php?id=11817 by Flugkiller 18:30:06 !crashlog 18:30:21 20527. Flugkiller, XL25 GnAM, T:92639 (milestone): https://underhound.eu/crawl/morgue/Flugkiller/crash-Flugkiller-20190104-225523.txt 19:03:06 amalloy: based on reply times for certain queries my conjecture is that the database doesn't have an index on explbr 19:03:10 !kw current 19:03:11 Keyword: current => cv>=0.22 19:03:12 !kw trunk 19:03:14 Keyword: trunk => alpha !experimental 19:03:22 this doesn't account for 100% of slowness 19:03:23 but 19:03:26 !kw alpha 19:03:27 Built-in: alpha => alpha=true 19:03:31 !kw experimental 19:03:32 Keyword: experimental => explbr!=||file~~(smithgod|jump|combo_god) 19:03:52 this turns otherwise index-using queries into ones that need a full table scan is my guess 19:04:05 I haven't actually asked Postgres to EXPLAIN 19:05:17 ebering: explbr doesn't get involved in simple queries like !lm * recent, right? 19:06:57 correct 19:07:02 well 19:07:05 !kw recent 19:07:05 Keyword: recent => cv>=0.21 19:07:08 yes 19:08:22 what bugs me is that even !lm * recent takes over 90s, which should just be a simple count(*) on an indexed column (cv) 19:08:47 er, not count(*) i guess. count(cv >= 0.21) 19:10:16 I'm not super familiar with sequell's sequell->sql compiler 19:10:47 yeah the issue is not just explbr; I take pains to use e.g. 'current alpha` all the time (or often just current) and timeouts are frequent 19:10:58 but I think !lm * recent compiles to a select * from milestones where cv>=0.21 order by time 19:11:06 I think there's not any particularly recent slowdown 19:11:38 but maybe a year or so ago snark seemed to changed something in sequell or something in the shallot.org config 19:11:55 I noticed that the timeout period increased around this same time when queries seemed like they were systematically slower 19:12:14 s/shallot.org/shalott.org/ 19:13:18 it's possible that something still more recent has changed but I've been seeing sequell responding slowly to !lm queries especially for well over a year even if you're careful to use just current or something similarly restrictive 19:14:21 the timeouts for something simple like `!lm * recent' also go back a ways but I'm not sure if it's gotten markedly worse more recently 19:15:00 given that we don't run shalott.org nor have access to its configuration, not sure there's a ton we can do about it; perhaps if you could set up a copy that had all the same data shalott.org did 19:17:51 the instructions in the repo include downloading all the milestones and logfiles, so in principle it wouldn't be hard to set up another copy. i've actually been waiting like 5 hours while it continues to fill up this giant database it made. who does run shalott? if we found and implemented a fix could we ask them to apply it? 19:18:48 green/snark does and probably unles the fix is "lift resource limits on postgres" 19:20:07 shalott.org is owned and operated by snark, yeah; you would probably want to directly discuss any low-level change with him 19:20:33 shalott.org automatically pulls in yml configuration changes, but for changes to sequell itself he'd probably have to apply those 19:20:49 https://github.com/crawl/sequell/commit/537c66359d1b5c6a3ad868c475c0529375355207 19:20:57 this is around when he raised the command timeout 19:21:40 doesn't appear any major changes happened proximal to that 19:23:44 snark has described shalott.org as "creaky" to me. depending on the nature of what you change and how that affects load average on the server, it could be good to have him take a look at any changes you propose 20:23:40 Guys, tossing potions seems to take into consideration throwing skill 20:24:11 Thrown potion on my Gn with 8 Throwing took 8 auts. 20:24:43 Usually items that are not designed for throwing always take 15 auts to throw. 20:37:37 Yermak: A quick test in wizmode (in vanilla trunk) takes 15 auts with 0 Throwing and 27; can you provide a save? 20:56:57 -!- amalloy is now known as amalloy_ 21:00:09 i see phantom orcs in clouds now 21:00:22 its going to transfer to dreams tonight, i know it 21:27:20 -!- amalloy_ is now known as amalloy 21:36:05 Pinkbeast, I tried it again just now, and it's 15. Before that I threw degeneration and I thought it was some kind of stalker's legacy. 21:36:56 I tested in wizmode with degeneration and it's 15 too. It looks like I made a mistake when observing 8 auts. 22:12:28 Yermak: No harm done. 22:32:30 New branch created: pull/936 (1 commit) 13https://github.com/crawl/crawl/pull/936 22:32:30 03Umer Shaikh02 07https://github.com/crawl/crawl/pull/936 * 0.23-a0-651-g5cf40e3: Fix information leak for invisible monsters 10(35 minutes ago, 1 file, 8+ 3-) 13https://github.com/crawl/crawl/commit/5cf40e3fdc2e 22:38:10 -!- muffindrake1 is now known as muffindrake