[mono2018-02]System test failures observed on device

[GouriKumari]

Steps to Reproduce

1. Update System to Xamarin.iOS from mono2018-02 branch d458533
2. Execute System test on device

Expected Behavior

Tests should pass successfully

Actual Behavior

[FAIL] HttpRequestStreamTest.Post_EmptyRequestStream : #1
[FAIL] HttpResponseStreamTest.AutomaticDecompression : System.Net.WebException : Error getting response stream (ReadAsync): ReceiveFailure Unable to read data from the transport connection: Connection reset by peer.
[FAIL] WebClientTest.GetWebRequestOverriding : System.Net.WebException : The operation has timed out.

Environment

Xamarin.iOS from mono2018-02 branch d458533

Build Logs

Build Log: http://xqa.blob.core.windows.net/gist/report-a8c0c5d2cd7141c2bdc6f3640f6be371.txt
Test Log: http://xqa.blob.core.windows.net/gist/System-ios03-58-1587cf18d9f45ef54e0c940cda510cd49c97afeb9a03-58-15-3fe18bc47c024158b7c55eb9968b52b9.zip

Additional Info:

I could reproduce only HttpResponseStreamTest.AutomaticDecompression failure locally .
This issue is also observed on iPhoneX device: http://xqa.blob.core.windows.net/gist/report-7c46777a2fa941bea744213995615b12.txt

[lewurm]

can't reproduce it on an iPhone7. Could this be network related?

[GouriKumari]

I can reproduce only HttpResponseStreamTest.AutomaticDecompression locally. Both the device and my test machine are on the same network. This is the same setting for bots and devices as well. I did run the tests twice from bots and HttpResponseStreamTest.AutomaticDecompression fails consistently in every test run.

Build Log: https://gist.github.com/GouriKumari/0f749983ae91847b795783a53b9c04ba
Test Log: https://gist.github.com/GouriKumari/c433688bfbda8d3ac05265a76df39026

[lewurm]

I can reproduce HttpResponseStreamTest.AutomaticDecompression now too. It isn't 100% reproducible for me, but fairly likely.

[lewurm]

@baulig @marek-safar any idea what this could be about?

[marek-safar]

This could be caused by @baulig async changes. Martin could you confirm?

[lewurm]

comment by @baulig https://xamarinhq.slack.com/archives/C5P3G23NG/p1522789572000010

[baulig]

I don't think people on github can read the slack archives.

What I said was essentially that GZip / Deflate decompression should not be considered "fully supported" (as in me spending time on it) until the new code that I'm currently working on - which will almost completely rework the way how GZip / Deflate decompression is handled - has landed. I simply do not consider it a worthy investment of my time to investigate GZip-related bugs against the old code and would rather ensure they are working with the new one.

The other two issues seem unrelated to this, though.

Time-wise, I expected the new GZip code to land by the end of this week, but this has now been delayed because I got distracted working on #7356.

Could we possibly schedule this issue as a bugs-week issue for next week? If possible, then I would like to continue investigating #7356 first as it seems like we're having a really major memory leak.

[GouriKumari]

The above mentioned tests are no longer included in mono2018-02 test run, hence there are no failures related to above tests. There is a random SocketTest failure reported on a single device, [FAIL] SocketTest.SocketErrorTest : SocketError wait timed out which seems to be network related.

Logs:

http://xqa.blob.core.windows.net/gist/System-ios01-10-33e978c67571aa0652bdb4ad6b8338ec46633e07ef01-10-33-e2f11dd682c3425cabc364fdd5471284.zip

Can someone confirm if this is a network related issue?
[Update] Not specific to mono2018-02, this issue occured on a device that executed d15-7 tests.
http://xqa.blob.core.windows.net/gist/report-683fdf6e957843c59f124c9614bf0add.txt

[dalexsoto]

@GouriKumari this does seem to be a network related issue since I could not reproduce this using 15.7 in device nor using simulator.

[GouriKumari]

I am closing this based on the above comment.