|
01:19
melezhik_ joined
01:21
melezhik_ left
01:33
melezhik_ joined
01:38
melezhik_ left
01:59
librasteve_ left
03:33
melezhik_ joined
03:35
melezhik_ left
04:06
melezhik_ joined
04:10
melezhik_ left
04:34
melezhik_ joined
04:36
melezhik_ left
07:36
melezhik_ joined
07:42
melezhik_ left,
melezhik_ joined
07:46
melezhik_ left
|
|||
| patrickb | I'm confused and seek clarification. | 08:12 | |
| github.com/rakudo/rakudo/blob/main...akumod#L33 | |||
| When looking at a META6.json file `provides` contains a hash where each value is a path string. The linked code seems to try to treat that string as a hash. What do I miss? | 08:14 | ||
|
09:13
melezhik_ joined
|
|||
| melezhik_ | . | 09:13 | |
|
09:20
melezhik_ left
09:23
melezhik_ joined
|
|||
| lizmat | ugexe ^^ | 09:26 | |
|
09:42
melezhik_ left
09:46
melezhik_ joined
09:52
melezhik_ left
10:10
librasteve_ joined
11:59
melezhik_ joined
|
|||
| melezhik_ | patrickb: [Coke]: can we run another round ? whenever you are ready (need to upgrade agent to the latest commit) | 12:04 | |
|
12:08
melezhik_ left
|
|||
| patrickb | booting up | 12:10 | |
|
13:00
melezhik_ joined
|
|||
| melezhik_ | ok, tnanks! | 13:00 | |
| started 2000 modules for Rakudo 2025.11 | 13:01 | ||
| patrickb: is agent running ? don't see heartbeats from it ... | 13:04 | ||
| ok, now I see | 13:06 | ||
| heartbeat from gallant-hellman-03405561, version: 0.0.38, max threads: 19, jobs run cnt: 0 | |||
|
13:13
melezhik_ left,
melezhik_ joined
13:25
melezhik_ left,
melezhik_ joined
|
|||
| ugexe | patrickb look at the meta6.json file that is installed. You’ll see that the provided section format is changed in CURI. I opened a PR to not do that many years ago but lost interest github.com/rakudo/rakudo/pull/2778 | 13:27 | |
| patrickb | Ah, that's good info. Thanks! | 13:30 | |
| btw. I'm currently digging into github.com/rakudo/rakudo/pull/5060 again. | 13:31 | ||
| I'll probably build the PR up from scratch again. | 13:32 | ||
|
13:43
melezhik_ left,
melezhik_ joined
13:50
melezhik_ left,
melezhik_ joined
|
|||
| melezhik_ | patrickb: ooo - did you cleanup agent container data ? the speed is pretty impressive | 13:59 | |
| ||| TESTS STAT: time: 57m | tests total: 2000 | finished tests: 1014 | |||
| patrickb | I just did a ./agent/run.sh | 14:00 | |
| melezhik_ | ah, ok, that means you completely cleaned cache, that's ok ) | 14:01 | |
| I guess you run patched run.sh? as the one in source is `docker run -d --rm --name agent -e BRW_AGENT_MAX_THREADS=1 -e BRW_AGENT_CAP_INSTALL_FROM_SOURCE_FORCE=1 -p 4002:4000 agent` | 14:02 | ||
| setting BRW_AGENT_MAX_THREADS to 1 | 14:03 | ||
| your one has 19 | 14:04 | ||
| btw, how can ask `zef install Foo` run tests in verbose mode? | 14:07 | ||
| showing every test line ? | |||
| test report / whatever | |||
|
14:17
melezhik_ left,
melezhik_ joined
|
|||
| patrickb | yes patched that file. threads 19 and s/docker/podman/ basically | 14:21 | |
| melezhik_ | ok! | 14:23 | |
|
14:47
melezhik_ left,
melezhik_ joined
14:56
melezhik_ left,
melezhik_ joined
15:04
melezhik_ left,
melezhik_ joined
15:24
melezhik_ left,
melezhik_ joined
|
|||
| melezhik_ | ||| TESTS STAT: time: 002h | tests total: 2000 | finished tests: 1998 | sent to queue: dist=0 / redist=2 | agents cnt: 1 | 15:25 | |
| finished | |||
| patrickb: I've just run another test for rakudo 2025.10, if you keep your agent running for about 2 hours - I'd appreciate it, thanks! | 15:27 | ||
| rightly it takes around 2 hours to test 2000 modules, on your 19 threads agent | |||
| roughly | 15:28 | ||
|
15:32
melezhik_ left
|
|||
| Geth | rakudo: ShimmerFairy++ created pull request #6034: Update Rakudo for Unicode 17.0.0 |
16:04 | |
| ab5tract | 2000 modules in 2 hours! Impressive :D | ||
| Geth | roast/unicode17-update: 5 commits pushed by Faye++ | 16:14 | |
| disbot6 | <melezhik.> Interesting . Still can’t get why Readljne tests fails for 2025.11 and pass for 2025.10? | 16:50 | |
|
16:51
melezhik joined
|
|||
| melezhik | Interesting . Still can’t get why Redline tests fails for 2025.11 and pass for 2025.10? | 16:51 | |
| brw.sparrowhub.io/file_view/brw-rep...adline.log | 16:52 | ||
| Actually this is a list for all found discrepancies between 2025.11 and 2025.10 ( round still running ) | 16:54 | ||
| brw.sparrowhub.io/report/brw-reporter/2587 | |||
| In artifacts one can find logs for failed modules in 2025.11, with successful tests info for 2025.10 | 16:55 | ||
| Another interesting finding is that brw.sparrowhub.io/file_view/brw-rep...Number.log | 16:56 | ||
| Intl::Format::Number | |||
| And Doublephone - brw.sparrowhub.io/file_view/brw-rep...ephone.log | 16:57 | ||
| [Coke] | Kicking off a new build/run now | 17:38 | |
| - JSONArgsRecommended: JSON arguments recommended for ENTRYPOINT to prevent unintended behavior related to OS signals (line 39) | 17:40 | ||
|
18:03
melezhik_ joined
18:07
melezhik_ left
|
|||
| melezhik | Ok. Looks like “ParaSeq” got stuck on agents , what’s wrong with module ? ) | 18:14 | |
| I guess some of the agents , maybe both now testing ParaSeq and it takes awhile , other jobs are just redundant, 1999 of 2000 finished | 18:19 | ||
| That’s this module that hold things now | 18:20 | ||
| usercontent.irccloud-cdn.com/file/...786061.JPG | 18:21 | ||
| [Coke] | nothing in the container really consuming cpu or memory at this point | 18:26 | |
| ... oh. looks like all 4 running jobs are ParaSeq? | |||
| each run that's the only thing on the list. | 18:27 | ||
| "don't do that to an agent". :) | |||
| melezhik | So your agent run "zef install ParaSec”? Right ? | 18:32 | |
| Can you bash to agent and do ps aux | grep zef ? | |||
| patrickb: what about your agent ? | 18:33 | ||
| It runs 9 jobs now I wonder what they are ? | |||
| So ParaSec is a “tough guy”? | 18:34 | ||
| [Coke] | ps aux | grep zef | wc -l == 25 | 18:35 | |
| melezhik | I need module name | ||
| Just “ps aux | grep zef” | 18:36 | ||
| Or you can look at any agent.job report via Ui at localhost | |||
| 4 running jobs | 18:38 | ||
| brw.sparrowhub.io/file_view/agent.2...r.data.txt | |||
| [Coke] | melezhik: it's 25 lines. | 18:40 | |
| melezhik | Can you please paste them in gists? | ||
| [Coke] | grep 'zef ' shows nothing. | ||
| it's got 2 dozen lines of (e.g.) /tmp/whateverable/rakudo-moar/3f8de27bcb2ee987a2997de8fe314c460b7e53f9/bin/rakudo -I CompUnit::Repository::Staging#name(site)#/tmp/.zef.1764785189.17261/1764785196.17261.1551.587401418667 t/11-reduce.rakutest | 18:41 | ||
| melezhik | Ok, it could unit tests | ||
| Ok we got something | |||
| t/11-reduce.rakutest | |||
| github.com/lizmat/ParaSeq/blob/mai...e.rakutest | 18:42 | ||
| My assumption that in 2025.10 ParaSec tests is extremely slow | 18:43 | ||
| Can you please grep by “ps aux | grep “whateverable/rakudo-moar” ? | 18:44 | ||
| Maybe we need to add it to skip list | 18:45 | ||
| It’d be interesting to see if we get the same processes on patrickb agent right now | 18:46 | ||
| [Coke] | there are 4 copies of that. | 18:48 | |
| corresponding to the 4 jobs I mentioed from the UI | |||
| melezhik | All 4 jobs ( pids ) have been running around 30 minutes | ||
| brw.sparrowhub.io/file_view/agent.2...r.data.txt | |||
| [Coke] | so the server should only have given me one copy, not 4. | ||
| melezhik | “9:48 PM <[Coke]> there are 4 copies of that.” This is expected as o10r resending ParaSec module test on all agents/threads | 18:49 | |
| Those are 4 different jobs running the same ParaSec module test and this is ok | 18:50 | ||
| [Coke] | ... you shouldn't send duplicate jobs to the *same* agent. | ||
| No, it's not. | |||
| I mean, it may be expected with the current setup, but if you have one job on the list, and I ask for 4 jobs.. don't give me 4 copies of the same job. | |||
| btw, you can get verbose test output by doing 'zef test --verbose .' - and separate that from the 'zef install', and then do a zef install with no tests. | 18:51 | ||
| Looks like ParaSeq tests are hanging (running on my laptop directly with verbose) | 18:52 | ||
| t/11-reduce.rakutest hangs after test 3/4 | |||
| melezhik | It’s arguable cause o10r logic does not rely on the fact that job has been already send to an agent , as for example job might have been finished but due to network partition results did not come back , so on my now it’s not reasonable to rely on just the fact that job has been already send as we don’t know where it’s finished or not | ||
| [Coke] | github.com/lizmat/ParaSeq/issues/4 | 18:53 | |
| ... if there are 2 different agents, that makes sense. | |||
| having me run two copies of the identical job in the same docker container is not. | |||
| melezhik | Ok . Anyways it looks it’s still an issue with 2025.10 | ||
| [Coke] | Given that paraseq issue, please add it to a skiplist. | ||
| melezhik | Can now please go to UI and terminate jobs ? | 18:54 | |
| Or through kill pid ? | |||
| “9:53 PM <[Coke]> Given that paraseq issue, please add it to a skiplist.” Yep | |||
| [Coke] | how long does it take to terminate? | 18:55 | |
| melezhik | Few seconds | ||
| [Coke] | clicked on each job, clicked terminate, no update when refreshing list of jobs. | ||
| melezhik | Usually needs few secs | ||
| [Coke] | Termination started: undefined | ||
| melezhik | It’s say operation | ||
| Ok that means you need to login | 18:56 | ||
| admin / admin | |||
| Login button on the left top | |||
| [Coke] | all dead | 18:57 | |
|
18:59
melezhik_ joined
|
|||
| melezhik_ | brw.sparrowhub.io/report/brw-reporter/2735 - diff 2025.11 VS 2025.10 | 19:04 | |
| [Coke]: ++ thanks | 19:05 | ||
| [Coke] | lizmat: updated that ticket - we're testing 200K elements - it works with 16, fails with 17 (not K, just 17) | ||
| s/fails/hangs/ | 19:06 | ||
| melezhik_ | meanwhile - github.com/melezhik/brownie/commit...ed67af2226 | 19:07 | |
| started new round for 2025.08 | 19:08 | ||
| on 2 agents ( 4 threads and 19 threads ) so it's going to be fast | |||
| lizmat | [Coke]: thanks! | 19:15 | |
| [Coke] | left that last one running for like 10m, nothing | ||
| melezhik_: yup, my agent is busy again. | 19:16 | ||
| melezhik_ | the last one ? | ||
| yeah, it's 2025.08 tests | |||
| for example your agent already send this - brw.sparrowhub.io/report/job.2.report/2991 | 19:17 | ||
| ||| TESTS STAT: time: 16m | tests total: 2000 | finished tests: 259 | sent to queue: dist=20 / redist=0 | agents cnt: 2 | 19:18 | ||
| [Coke] | I'll leave the laptop plugged in open. | ||
| "AND open" | 19:19 | ||
| patrickb | melezhik_: See my messages on 1st of December. ParaSeq and t/11-reduce.rakutest are known offenders. | 19:20 | |
| melezhik_ | ok ) we know this now )) | ||
| [Coke]: thanks for your help ! | 19:21 | ||
| [Coke] | guessing we never skipped it on blin, but I'm sure it took up a slot for testing, slowing everything down. | 19:23 | |
|
19:27
melezhik_ left,
melezhik_ joined
19:32
melezhik_ left,
melezhik_ joined
19:48
melezhik_ left
19:49
melezhik_ joined
19:56
librasteve_ left
19:58
melezhik_ left
19:59
melezhik_ joined
20:05
melezhik_ left,
melezhik_ joined
|
|||
| Geth | rakudo: 0rir++ created pull request #6035: Test .from-posix versus .from-posix-nanos |
20:19 | |
|
20:42
melezhik_ left,
melezhik_ joined
|
|||
| disbot6 | <melezhik.> brw.sparrowhub.io/file_view/brw-rep...adline.log | 20:49 | |
|
21:08
melezhik_ left
21:09
melezhik_ joined
21:15
melezhik_ left,
melezhik_ joined
|
|||
| melezhik_ | 2025.08 finished in hours 2.15 minutes | 21:36 | |
|
21:50
melezhik_ left
|
|||
| [Coke] | hour many hours? | 22:09 | |
| How many hours? | 22:10 | ||
|
22:46
melezhik_ joined,
melezhik_ left
23:43
rakkable left,
rakkable joined
23:46
melezhik left
23:47
melezhik_ joined
23:49
melezhik_ left
|
|||
