-
Notifications
You must be signed in to change notification settings - Fork 5k
[HttpListener] Hang of System.Net.HttpListener.Tests in CI #21940
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
Comments
fyi @hughbe @stephentoub note these are in master. |
Questions:
|
If we see this actively hung in any CI jobs, please let me know and I will try to debug it. Bonus points to anyone who logs into a machine and attaches a debugger before the process terminates 😄 |
Managed to catch one in CI. At least in the one I saw, it appears to be hanging here: @geoffkizer, looks like this might be the same hang you're investigating locally... |
Yep, that's exactly what I see locally. |
@stephentoub @geoffkizer does this mean if we disable Read_FullLengthAsynchronous_Success it will avoid this hang? If so could one of you please disable it? |
I don't think so; I think there are several (not sure how many) tests that hit the same underlying issue or issues. I don't believe it's a problem with a specific test, but rather with something in the implementation (or something systemic to many tests via code that's shared between them). It's possible we could find and disable enough tests to avoid the hangs. @geoffkizer has been investigating... |
But I'll see if I can temporarily disable enough to minimize the impact... |
A bunch of tests have been temporarily disabled. Just search for 20246 to find them. Let's try to get to the bottom of this asap. Hopefully I disabled enough to prevent the hangs in CI. |
Helix run failed again but I don't know whether it contained that change. |
@mellinoe was there any progress on tracking down the root cause of the hangs? |
@karelz I think @stephentoub had caught this (or a related test) under a debugger at one point. I was not able to successfully get a local repro despite running the tests in a loop for a long time. |
@geoffkizer and I fixed the issues we knew about. It's possible there's another lurking. |
Might be worth running the tests in a loop on macOS for a while and seeing if we can get something to hang. |
I borrowed a MacBook and will try running the tests in a loop here, once I've got my environment sync'd up. |
This is now the most impactful of all corefx test issues. |
@stephentoub and I spent several hours debugging this earlier today, and we weren't able to identify the exact problem yet. We saw some strange things that made us doubt the validity of the info coming from lldb/SOS, and ended up stumped for now. At the very least, the MacBook I'm using is able to reproduce the issue almost every single time I run the tests. So further investigations should be easy to do. |
This issue is repro on Core Tests (2.0.0), detail: https://mc.dot.net/#/product/netcore/200/source/official~2Fcorefx~2Frelease~2F2.0.0~2F/type/test~2Ffunctional~2Fcli~2F/build/20170605.01/workItem/System.Net.HttpListener.Tests/wilogs |
Failed again on Core Tests (Master) in build 20170607.01, detail: https://mc.dot.net/#/product/netcore/master/source/official~2Fcorefx~2Fmaster~2F/type/test~2Ffunctional~2Fcli~2F/build/20170607.01/workItem/System.Net.HttpListener.Tests/wilogs |
@mellinoe Can you share the repro that you used for this.. Did you run the httplistener tests in loop or full test run? Were you able to identify the test that's hanging? |
@Priya91 There are several tests that hang. It repros on your laptop every time, just doing a regular run of the test project by itself. |
This issue is repro on Core Tests (2.0.0), detail: https://mc.dot.net/#/product/netcore/200/source/official~2Fcorefx~2Frelease~2F2.0.0~2F/type/test~2Ffunctional~2Fcli~2F/build/20170615.01/workItem/System.Net.HttpListener.Tests/wilogs |
Need an update from @mellinoe who has been investigating this. |
I've debugged through a few repros at this point, but I haven't identified a root cause. One repro was hung in this method, https://github.com/dotnet/corefx/blob/master/src/System.Net.HttpListener/src/System/Net/Managed/HttpListenerRequest.Managed.cs#L302, in what could potentially be an infinite loop. That's the closest we've gotten to the bottom of this. |
Per @karelz moving to 2.0 |
This comment has been minimized.
This comment has been minimized.
I think the last association is incorrect: [UPDATE] The referenced comment was hidden as off-topic by @karelz
This seems to be clearly environmental issues. |
Closing per @wfurt's analysis above. |
There are two tests disabled against this issue: runtime/src/libraries/System.Net.HttpListener/tests/HttpResponseStreamTests.cs Lines 380 to 385 in 04f2226
runtime/src/libraries/System.Net.HttpListener/tests/HttpResponseStreamTests.cs Lines 419 to 424 in 04f2226
|
https://ci.dot.net/job/dotnet_corefx/job/master/job/ubuntu14.04_debug_prtest/8257/consoleFull#130661499679494335-f7bd-47d0-8771-8661e00c2db2
Not actionable based on this information, but starting an issue.
The text was updated successfully, but these errors were encountered: