-
Notifications
You must be signed in to change notification settings - Fork 29.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
test: remove custom rimraf #30074
test: remove custom rimraf #30074
Conversation
Overlap with #29235. |
Can't clean tmpdir: c:\workspace\node-test-binary-windows-2\test\.tmp.124
Files blocking: [ 'node-copy.exe' ]
c:\workspace\node-test-binary-windows-2\test\common\tmpdir.js:57
throw e;
^
Error: EPERM: operation not permitted, unlink '\\?\c:\workspace\node-test-binary-windows-2\test\.tmp.124\node-copy.exe'
at unlinkSync (fs.js:1051:3)
at fixWinEPERMSync (internal/fs/rimraf.js:248:5)
at rimrafSync (internal/fs/rimraf.js:187:14)
at internal/fs/rimraf.js:211:9
at Array.forEach (<anonymous>)
at _rmdirSync (internal/fs/rimraf.js:210:25)
at fixWinEPERMSync (internal/fs/rimraf.js:246:5)
at rimrafSync (internal/fs/rimraf.js:187:14)
at Object.rmdirSync (fs.js:763:12)
at process.onexit (c:\workspace\node-test-binary-windows-2\test\common\tmpdir.js:42:8) {
errno: -4048,
syscall: 'unlink',
code: 'EPERM',
path: '\\\\?\\c:\\workspace\\node-test-binary-windows-2\\test\\.tmp.124\\node-copy.exe'
} |
I was afraid this would happen on Windows.... |
@Trott great work on this, it would be exciting to dog-food our own rimraf, I'm fairly far along adding coverage for Windows to our windows builds, which might help us better track down the edge-cases that are breaking this build. |
rimrafSync(this.path, opts); | ||
function refresh() { | ||
try { | ||
fs.rmdirSync(this.path, { recursive: true }); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@Trott could setting maxBusyTries
here (and other call sites) help the issue with Windows in the CI?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I thought I tried that and it seemed like maybe it was ignored in the synchronous version?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yep. It looks like all of the retry logic is in the async version.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Is the solution to pull the retry logic into the sync version as well? @isaacs was there a reason it only exists in the async
version?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
So it turns out there is some hard coded retry logic for the sync version:
node/lib/internal/fs/rimraf.js
Lines 214 to 218 in f171112
for (let i = 0; i < numRetries; i++) { | |
try { | |
return rmdirSync(path, options); | |
} catch {} // Ignore errors. | |
} |
setTimeout()
(Node would need a sleep API I think).
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Could we move some of the sync implementation into C and use usleep
, or just expose sleep for internal use only?
I'm guessing it's an intentional design decision that Node has no sleep
?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Two ideas, but I think they're both problematic:
-
We do have
common.busyLoop()
so if we put some retry logic incommon/tmpdir.js
, that could be used. Seems like a suboptimal hack, though. -
Since this is all happening on cleanup, we could also decide to remove the cleanup functionality. The upside is that if someone wants to check something about the tmp files in an
exit
handler, they don't have to worry about anotherexit
handler removing the tmp files before they get a chance to do the check. The downside, of course, is that we'll have all these tmp directories left lying around.
I don't think either of those are the way to go, but if they help anyone else arrive at a better idea, cool.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm guessing it's an intentional design decision that Node has no sleep?
I don't know if it's intentional or not. libuv has a uv_sleep()
function in its test runner. I can look into exposing it. I think sleep()
is a reasonable thing for Node to have outside of this use case.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@cjihrig I think this could be great 👍 and we start to gain some benefits from pulling rimraf
in to core 😄
I imagine #30569 will be good to land now and that this can be closed once it does. |
Use internal recursive rmdir() in the common/tmpdir module we use in tests.
Use internal recursive rmdir() in the common/tmpdir module we use in
tests.
Checklist
make -j4 test
(UNIX), orvcbuild test
(Windows) passes