build: spawn make test-ci with -j1#23733
Conversation
|
Do we know how much time a CI run takes with and without |
Good question (I don't see any significant change) with
|
| Job | Build # | Duration |
|---|---|---|
| node-compile-windows | build #21674 | ( 46 min ) |
| node-test-binary-windows | build #20858 | ( 13 min ) |
| node-test-commit | build #22448 | ( 1 hr 40 min ) |
| node-test-commit-arm | build #19316 | ( 39 min ) |
| node-test-commit-custom-suites-freestyle (test-worker) | build #1730 | ( 8 min 16 sec ) |
| node-test-commit-freebsd | build #21307 | ( 18 min ) |
| node-test-commit-linux | build #22483 | ( 1 hr 40 min ) |
| node-test-commit-linux-containered | build #7947 | ( 1 hr 39 min ) |
| node-test-commit-linuxone | build #6326 | ( 6 min 41 sec ) |
| node-test-commit-osx | build #22011 | ( 1 hr 3 min ) |
| node-test-commit-plinux | build #21033 | ( 8 min 32 sec ) |
| node-test-commit-smartos | build #21042 | ( 31 min ) |
| node-test-commit-windows-fanned | build #21620 | ( 1 hr 2 min ) |
| node-test-linter | build #22905 | ( 5 min 2 sec ) |
Without (daily master)
| Job | Build # | Duration |
|---|---|---|
| node-compile-windows | build #21670 | ( 45 min ) |
| node-test-binary-windows | build #20854 | ( 13 min ) |
| node-test-commit | build #22444 | ( 1 hr 23 min ) |
| node-test-commit-arm | build #19312 | ( 36 min ) |
| node-test-commit-custom-suites (test-internet) | build #724 | ( 1 min 59 sec ) |
| node-test-commit-custom-suites-freestyle (test-worker) | build #1725 | ( 9 min 5 sec ) |
| node-test-commit-freebsd | build #21303 | ( 17 min ) |
| node-test-commit-linux | build #22479 | ( 29 min ) |
| node-test-commit-linux-containered | build #7944 | ( 1 hr 22 min ) |
| node-test-commit-linuxone | build #6319 | ( 6 min 27 sec ) |
| node-test-commit-osx | build #22007 | ( 18 min ) |
| node-test-commit-plinux | build #21029 | ( 8 min 18 sec ) |
| node-test-commit-smartos | build #21038 | ( 27 min ) |
| node-test-commit-windows-fanned | build #21616 | ( 59 min ) |
| node-test-linter | build #22901 | ( 4 min 25 sec ) |
|
@refack any reason that this is not author ready? |
No CI since the most recent change? |
This need to be followed up in Jenkins, so mainly so this will not be landed by mistake. |
All the sub targets have internal parallelism, so no performance loss. Also `make` doesn't to a good enough job of combining the output streams, or eliminate races. PR-URL: nodejs#23733 Fixes: nodejs#22006 Reviewed-By: Richard Lau <riclau@uk.ibm.com> Reviewed-By: Joyee Cheung <joyeec9h3@gmail.com> Reviewed-By: Matheus Marchini <mat@mmarchini.me>
|
Since this allowed us to restore |
All the sub targets have internal parallelism, so no performance loss. Also `make` doesn't to a good enough job of combining the output streams, or eliminate races. PR-URL: #23733 Fixes: #22006 Reviewed-By: Richard Lau <riclau@uk.ibm.com> Reviewed-By: Joyee Cheung <joyeec9h3@gmail.com> Reviewed-By: Matheus Marchini <mat@mmarchini.me>
All the sub targets have internal parallelism, so no performance loss. Also `make` doesn't to a good enough job of combining the output streams, or eliminate races. PR-URL: #23733 Fixes: #22006 Reviewed-By: Richard Lau <riclau@uk.ibm.com> Reviewed-By: Joyee Cheung <joyeec9h3@gmail.com> Reviewed-By: Matheus Marchini <mat@mmarchini.me>
|
landed in 10.x in 2cd68be |
|
just catching up on this. Did we end up fixing the addon build make+stdout problems with this change? nice work if so! |
Yes, but we have a deeper issue. |
All the subtasks have internal parallelism, so there is minimal loss.
Also
makedoesn't do a good enough job of combining the outputstreams, or eliminate races.
Fixes: #22006
Ref: #23423 (comment)
Ref: #23286 (comment)
/CC @nodejs/build @nodejs/build-files
Checklist
make -j4 test(UNIX), orvcbuild test(Windows) passes