02:24:37  * theresajayne17joined
02:24:57  * theresajayne17quit (Read error: Connection reset by peer)
03:12:36  * ChickenSoup_joined
03:14:23  * ChickenSoup_quit (Remote host closed the connection)
03:15:30  * rosseaux14joined
03:17:04  * rosseaux14quit (Remote host closed the connection)
03:43:21  * Cprossu10joined
03:43:36  * Cprossu10quit (Remote host closed the connection)
04:16:57  * KobrAs12joined
04:17:07  * KobrAs12quit (Remote host closed the connection)
06:39:19  * node-ghjoined
06:39:19  * node-ghpart
06:55:14  * node-ghjoined
06:55:15  * node-ghpart
09:18:11  * Nojoined
09:18:30  * Noquit (Remote host closed the connection)
09:24:11  * VampiricPadraigjoined
09:24:22  * VampiricPadraigquit (Remote host closed the connection)
09:26:24  * myth0d15joined
09:26:34  * myth0d15quit (Remote host closed the connection)
09:33:15  * lolmacjoined
09:33:21  * lolmacquit (Remote host closed the connection)
09:39:07  * node-ghjoined
09:39:07  * node-ghpart
09:42:54  * Holo28joined
09:43:09  * Holo28quit (Remote host closed the connection)
09:58:22  * Exagone31319joined
10:00:14  * Exagone31319quit (Remote host closed the connection)
10:03:26  * supercool20joined
10:05:19  * supercool20quit (Remote host closed the connection)
10:07:30  * pppingme12joined
10:09:23  * pppingme12quit (Remote host closed the connection)
10:13:34  * node-ghjoined
10:13:35  * node-ghpart
10:20:08  * RyanKnack1joined
10:21:55  * RyanKnack1quit (Remote host closed the connection)
10:33:37  * theresajayne10joined
10:35:30  * theresajayne10quit (Remote host closed the connection)
11:14:55  * mz`8joined
11:16:48  * mz`8quit (Remote host closed the connection)
11:26:26  * Geeky_Bearjoined
11:28:18  * Geeky_Bearquit (Remote host closed the connection)
11:44:30  * node-ghjoined
11:44:30  * node-ghpart
11:45:13  * node-ghjoined
11:45:14  * node-ghpart
11:46:39  * node-ghjoined
11:46:40  * node-ghpart
12:01:04  * node-ghjoined
12:01:05  * node-ghpart
12:16:00  * mappumjoined
12:17:54  * mappumquit (Remote host closed the connection)
13:34:01  * ExeciN25joined
13:35:53  * ExeciN25quit (Remote host closed the connection)
15:11:45  * ugjka27joined
15:14:29  * ugjka27quit (Remote host closed the connection)
15:23:33  * Sharkerjoined
15:24:25  * Sharkerquit (Remote host closed the connection)
15:42:13  <joyee>freebsd11 has been failing `addons/stringbytes-external-exceed-max/test-stringbytes-external-exceed-max-by-1-binary` constantly (all just `crashed`, memory issue?)
15:42:23  <joyee>https://github.com/nodejs/reliability/issues/12
15:44:55  <joyee>Also, all of them are digital ocean ones
15:50:01  <mmarchini>Uh, I thought those were just flakes. I'll check once I get back to the office
15:50:10  <mmarchini>But memory issue sounds reasonable
15:51:08  <mmarchini>Maybe we should take these nodes offline?
15:54:34  <joyee>Maybe wait for a while until we figure it out, it failed the lastest 17/36 runs
15:57:07  <joyee>Maybe it's because os.totalmem() does not work on those digital ocean freebsd machines?
15:58:09  <refack>They were offline for a few days. So it could be bit-rot
15:59:22  <joyee>They are re-enabled in https://github.com/nodejs/build/issues/1441
15:59:43  <joyee>Maybe something was missing when the images were rebuilt?
16:01:32  * node-ghjoined
16:01:32  * node-ghpart
16:01:42  * node-ghjoined
16:01:42  * node-ghpart
16:10:30  <refack>Oh poo. `tap2junit -i test.tap` doesn't pick those as failed tests.
16:19:39  <refack>totalmem works:
16:19:45  <refack>https://www.irccloud.com/pastebin/eG7hAF0H/
16:20:04  <refack>and 2GB is not small
16:32:08  <refack>ohh, but
16:32:12  <refack>https://www.irccloud.com/pastebin/IJY8Zwwn/
16:51:28  <joyee>hm, wait, why isn't common.enoughTestMem using freemem?
16:51:48  <refack>my tought as well
16:52:24  <refack>I guess since it's a transient parameter, while `totalmem` is a platform constant
16:53:03  <joyee>Still it should've throw proper errors instead of outright crashing..
16:53:14  <refack>I assume we want to always skip this test on certain platforms. not being dependant on transient conditions
16:53:20  <refack>Yes
16:53:36  <refack>AFAICT the test does check if it can allocated that amount of memry
16:53:53  <refack>https://www.irccloud.com/pastebin/1rZimPYc/
16:57:15  <joyee>It's also strange that only the latin-1 encoding test failed
16:57:24  <joyee>Other encodings worked fine
16:57:34  <refack>yes
17:06:06  * ktrjoined
17:09:30  * ktrquit (K-Lined)
17:09:54  * benjamincoejoined
17:13:15  <joyee>I can reproduce the failure on that machine, I'll see if I can get a coredump out
17:13:28  <refack>I'm doing that as well
17:13:37  <refack>It seems like the test _doesn't_ fail
17:13:54  <refack>i.e. `buf.tostring()` completes
17:15:33  <joyee>https://www.irccloud.com/pastebin/xTR3JET2/
17:15:45  <joyee>I've gone one repro, but didn't save the core dump
17:16:10  <refack>https://www.irccloud.com/pastebin/cnsTLEXA/
17:16:28  * benjamincoequit
17:16:46  <joyee>That didn't fail with signal 9 it seems
17:17:07  <refack>ohh right
17:17:27  <refack>But I think I understand why this test is special
17:17:52  <refack>besides the overflowing `buf.tostring()`
17:18:19  <refack>it also does `let maxString = buf.toString('latin1', 1);`
17:18:46  <refack>so at that point I assume the process is thrashing to the swap file
17:19:01  <refack>ohh and a third `maxString = buf.toString('latin1', 0, kStringMaxLength);`
17:20:24  <refack>Seems right
17:20:28  <refack>https://www.irccloud.com/pastebin/vzexNAXk/
17:20:39  <joyee>It tried to let the GC collect that string
17:20:40  <refack>third allocation fails
17:22:18  <refack>maybe split to 3 tests. or add an async step or an explicit GC call
17:22:48  <joyee>I am not even sure `maxString = undefined;` does what it's supposed to do
17:23:03  <refack>It's just a hint
17:23:13  <refack>You still need the GC to run
17:23:28  <joyee>Yeah, I am not sure if the GC actually cares
17:23:52  <refack>From the timing of the log output it does seem like there is heavy swaping going on
17:24:18  <joyee>Can you run it with `--trace_gc --trace_gc_verbose`?
17:25:28  <refack>running.
17:25:33  <refack>I know you can:
17:25:41  <refack>https://www.irccloud.com/pastebin/qE2nPCG4/
17:26:16  <joyee>Yeah I'd use that for this particular test
17:26:59  <joyee>But then...if there isn't enough space, the test still shouldn't crash
17:27:25  <refack>That could be an OS bug
17:28:19  <refack>BTW when I run it directly I get `$? 137`
17:28:27  <refack>not -9
17:29:00  <refack>ohh that the same
17:29:55  <refack>I have one failing with trace
17:30:00  <refack>https://www.irccloud.com/pastebin/FI3zgCwR/
17:31:48  <refack>and one that passed
17:31:51  <refack>https://www.irccloud.com/pastebin/059SqSBt/
17:32:19  <refack>The GC doesn't run between sync JS lines
17:32:27  <refack>https://www.irccloud.com/pastebin/NwfgnCMA/
17:32:47  <refack>this is the code
17:32:57  <refack>https://www.irccloud.com/pastebin/nT0Ds9wg/test.js
17:34:21  <refack>https://www.irccloud.com/pastebin/Sxvuhjxd/
17:34:40  <refack>even the failed `toString` doesn't get released
17:35:09  <refack>BTW it's `External` memory. Does the GC even manage that?
17:35:49  * ajvpot6joined
17:39:27  <refack>last one, with explicit `global.gc()` calls
17:39:33  <refack>https://www.irccloud.com/pastebin/zMF1YpIa/
17:39:37  * ajvpot6quit (Remote host closed the connection)
17:41:28  * MatthewAllan9323joined
17:46:14  * MatthewAllan9323quit (Remote host closed the connection)
17:48:16  <joyee>GC should collect the JS strings which should be responsible of freeing those external memory
17:48:44  <refack>Right, thanks.
17:52:21  <joyee>So calling the v8 helper does the trick?
17:52:30  <refack>yes
17:53:13  <refack>But like you said, it should crash. So I'm trying to get a core
17:53:22  <refack>*should not
18:07:02  * szt29joined
18:08:25  * wolsjoined
18:09:35  * szt29quit (Remote host closed the connection)
18:12:50  * wolsquit (Remote host closed the connection)
18:20:51  * BridgeARjoined
18:29:21  * insidious28joined
18:31:21  * insidious28quit (Remote host closed the connection)
18:39:42  * BridgeARquit (Remote host closed the connection)
18:52:23  * BridgeARjoined
19:11:36  * node-ghjoined
19:11:36  * node-ghpart
19:16:54  * mmarchini_joined
19:17:30  * CC668joined
19:17:40  * zkat_joined
19:17:52  * ryzokuken_joined
19:18:35  * rvagg_joined
19:18:45  * ofrobots__joined
19:19:12  * CC668quit (Remote host closed the connection)
19:21:39  * BridgeARquit (Remote host closed the connection)
19:25:18  * ryzokukenquit (*.net *.split)
19:25:18  * zkatquit (*.net *.split)
19:25:25  * ofrobots_quit (*.net *.split)
19:25:25  * mmarchiniquit (*.net *.split)
19:25:26  * rvaggquit (*.net *.split)
19:25:29  * zkat_changed nick to zkat
19:25:30  * ryzokuken_changed nick to ryzokuken
19:25:30  * rvagg_changed nick to rvagg
19:43:46  <joyee>refack: any progress? It's still failing CIs, maybe we should just use the `global.gc` quick fix to make CI green again first
19:44:13  <refack>Sure. I'l write up a PR
20:08:32  * node-ghjoined
20:08:32  * node-ghpart
20:22:06  <refack>Here we go https://github.com/nodejs/node/pull/22301
20:26:30  * BridgeARjoined
20:38:12  * Dwarf17joined
20:41:08  * Dwarf17quit (Remote host closed the connection)
21:07:45  <Trott>Is this something I should log onto the relevant machine and clean up manually to see if it fixes the pipeline job? https://github.com/nodejs/node/pull/21870#issuecomment-412664493
21:08:59  <Trott>Looks like it's test-joyent-freebsd10-x64-2 so...FreeBSD, and I know that's been A Thing lately?
21:09:25  <refack>I think it's an actual github refresh bug
21:09:40  <Trott>Yeah, `git status` now says it's clean....
21:09:51  <Trott>(I might be on the wrong host. I'm looking at the most recent failure now.)
21:10:20  <Trott>Oh, this:
21:10:21  <Trott>https://www.irccloud.com/pastebin/GVNQ67qp/
21:11:02  <refack>When a test-pull-request starts and does a rebase it calculates the git SHA of the rebased branch. If one of the workers fetches a different git ref, that's the fail
21:11:35  <refack>that's the linter, right?
21:12:01  <Trott>Yes.
21:12:06  <Trott>https://ci.nodejs.org/job/node-linter/529/console
21:13:08  <Trott>Mildly more context if it helps:
21:13:09  <Trott>https://www.irccloud.com/pastebin/zVMNlk3K/
21:14:44  <Trott>Oh, that might be something that only happens when the linter fails? Still should be fixed of course.
21:25:54  <joyee>Shouldn't https://github.com/nodejs/build/blob/master/jenkins/scripts/node-test-commit-pre.sh do `git clean -fxd` somewhere?
21:26:22  <refack>The Jenkins git plugin does it
21:26:47  <joyee>hmm, it's strange that there have been a lot of unstaged files errors lately
21:27:00  <joyee>When does that plugin run?
21:27:48  <refack>at the start of each job, it's usually the only code that does the checkout
21:28:36  <refack>I think it happens when we have shell script that does rebasing
21:35:29  <joyee>https://ci.nodejs.org/job/node-test-commit-arm/17682/nodes=centos7-arm64-gcc6/console
21:35:50  <joyee>I can't figure out why there was untracked files with this log
21:36:07  <joyee>s/untracked files/unstaged changes/
21:37:09  <joyee>hmm..it's cloning the bash script from maclover7's repo?
21:38:03  * node-ghjoined
21:38:03  * node-ghpart
21:39:34  <refack>re: linter. there was an old PR with a fix waiting https://github.com/nodejs/build/pull/1349
21:40:39  <refack>The "Untracked" can be expected. the `modified: deps/v8/third_party/jinja2/LICENSE` is a weird thing I've been seing even localy
21:41:01  <refack>I'm assuming it's and EOL issue that git "auto-fixes"
21:45:46  <refack>This is how the linter looks now - https://ci.nodejs.org/blue/organizations/jenkins/node-linter/detail/node-linter/536/pipeline
21:52:35  * BridgeARquit (Ping timeout: 240 seconds)
21:55:16  <joyee>refack: ouch, that file is using CRLF
21:55:53  <refack>Don't get me started...
21:56:37  <joyee>Should we just work around this with .gitattribute?
21:57:09  <refack>I run (on windows) with `binary`
21:57:32  <refack>except `.bat` files, windows doesn't care anymore
21:58:13  <refack>For this I don't know. Ask targos
22:02:06  <joyee>V8 is currently using `* text=auto`
22:02:44  <refack>Then we need to wither fix that file. Or change .gitglobal on that machine...
22:03:29  <refack>`core.autocrlf = input`
22:05:38  <joyee>Maybe we should take another stab at https://github.com/nodejs/node/pull/7019
22:09:24  <refack>But we'll need to patch https://github.com/nodejs/node/blob/master/deps/v8/.gitattributes anyway
22:11:00  * Hinkjoined
22:12:11  * Hinkquit (Remote host closed the connection)
22:15:29  <joyee>um...does that file affect us?
22:15:51  <joyee>Git should not read it when we check out the node repo?
22:16:15  <refack>I assumed it's recursive like .gitignore
22:17:04  <joyee>If it's read by git, then when someone updates deps/v8, that file should've been converted to use LF before being checking into the repo right?
22:17:43  <joyee>hm, looks like the license file is older than the .gitattributes
22:18:45  <refack>yeah, I also need to read if `*` is like `*.*` or it might be special for `LICENSE` and `README` files
22:21:49  <refack>Actualy I would think that CI machines should run with `core.autocrlf = false`
22:22:06  <refack>They are not dev machines. They should use the files as is
22:22:23  <refack>like we use `npm ci`
22:28:07  * mmarchini_part
22:28:55  * mmarchinijoined
22:29:23  <mmarchini>Agreed
22:29:33  <mmarchini>BTW, yesterday I fixed this issue by removing the dirty workspaces
22:29:58  <mmarchini>I know this won't fix the issue in the long term, but should be enough to pass some CI runs
22:33:12  <joyee>mmarchini: that means theoratically we'd need to wipe every workspace so that the file is checked out properly from scratch?
22:45:03  <mmarchini>We might need to do that anyway after setting `core.autocrlf = false`, unless git is able to revert the EOL changes in these files after setting this option
22:46:59  <mmarchini>Do we know how many workspaces are affected? I found two yesterday
22:50:01  <joyee>See https://github.com/nodejs/reliability/issues/12
22:50:14  <joyee>test-packetnet-centos7-arm64-2, test-packetnet-centos7-arm64-1, test-digitalocean-ubuntu1604_sharedlibs_container-x64-7, test-digitalocean-ubuntu1604_sharedlibs_container-x64-10
22:57:55  <mmarchini>We could set `core.autocrlf = false` manually on these machines and try to `git checkout deps/v8/third_party/jinja2/LICENSE`. If that works we can write an ansible playbook to set this option on all servers
22:58:13  <refack>It's not working
22:58:18  <refack>not even rm
22:58:27  <refack>I thing I need to clear some cache
22:59:18  <mmarchini>Even after setting `core.autocrlf = false`?
23:03:09  <refack>yes. I remembered it being a pain
23:04:59  <refack>what a good way to see what the diff is really about
23:05:05  <refack>?
23:08:42  <refack>https://www.irccloud.com/pastebin/MeIaztbj/
23:08:45  <refack>help!
23:12:40  <refack>wait. I got it
23:14:35  <refack>ohh it broke again
23:14:39  <refack>https://www.irccloud.com/pastebin/sEVk3A9c/