Bug 33307 – websocket/tests all time out intermittently on Snow Leopard Release Bot (and possibly other bots)

• Component: Tools / Tests
• Status: NEW

[Eric Seidel (no email)]

websocket/tests/handshake-error.html timed out on Snow Leopard Release Bot

This is the first time I've noticed this, but I expect it's been flakey for a while:
http://build.webkit.org/results/SnowLeopard%20Intel%20Release%20(Tests)/r52898%20(3952)/results.html

I also suspect this probably has the same root cause as bug 33306.

[Eric Seidel (no email)]

websocket/tests/bad-sub-protocol.html just timed out on the Snow Leopard Bot:
http://build.webkit.org/results/SnowLeopard%20Intel%20Release%20(Tests)/r52903%20(3957)/results.html
I suspect this is the same root cause.

[Eric Seidel (no email)]

websocket/tests/cross-origin.html just timed out on Snow Leopard.  I suspect it's also the same root cause.
http://build.webkit.org/results/SnowLeopard%20Intel%20Release%20(Tests)/r57103%20(7709)/results.html

[Eric Seidel (no email)]

Changing title.  I'm going to dupe the other bugs against this one.  This is a problem for all websocket tests it appears.

[Eric Seidel (no email)]

*** Bug 33306 has been marked as a duplicate of this bug. ***

[Eric Seidel (no email)]

websocket/tests/handshake-fail-by-cross-origin.html too.

[Eric Seidel (no email)]

*** Bug 33291 has been marked as a duplicate of this bug. ***

[Eric Seidel (no email)]

websocket/tests/url-parsing.html too.

[Alexey Proskuryakov]

I'm getting a reproducible timeout on a debug build with 

run-webkit-tests websocket/tests/close-on-navigate-new-location.html --repeat-each 500

This doesn't happen on Leopard (a different machine with fewer cores though).

[Alexey Proskuryakov]

This is a client side problem - it also happens when WebSocket server runs on a Leopard machine (with appropriate changes to DRT and the test itself).

[Eric Seidel (no email)]

When you say "client side problem" I assume you mean a problem in WebKit or its dependent libraries?

[Alexey Proskuryakov]

I ran a brief test with IPv6 disabled via system configuration, and still got a hang with steps from comment 8. Independent confirmation would help though, because I have a number of local modifications to test runner at the moment.

> When you say "client side problem" I assume you mean a problem in WebKit or its
> dependent libraries?

Yes, that's what I meant.

[Eric Seidel (no email)]

websocket/tests/unicode.html -> timed out
websocket/tests/httponly-cookie.pl -> timed out
websocket/tests/handshake-fail-by-no-cr.html -> timed out
websocket/tests/frame-length-skip.html -> timed out
websocket/tests/frame-length-skip.html -> timed out
websocket/tests/frame-length-skip.html -> timed out
websocket/tests/frame-length-longer-than-buffer.html -> timed out
websocket/tests/cross-origin.html -> timed out
websocket/tests/close-on-unload-reference-in-parent.html -> timed out
websocket/tests/bufferedAmount-after-close.html -> timed out
websocket/tests/bad-sub-protocol-non-ascii.html -> timed out
websocket/tests/bad-sub-protocol-empty.html -> timed out
websocket/tests/bad-sub-protocol-empty.html -> timed out

Are timeout results from the last week from one of the Snow Leopard commit-queue bots.

websocket timeouts account for about half of all timeouts seen on the CQ.  The other half are http tests:

http/tests/xmlhttprequest/access-control-basic-non-simple-allow-async.html -> timed out
http/tests/xmlhttprequest/access-control-basic-non-simple-allow-async.html -> timed out
http/tests/plugins/interrupted-get-url.html -> timed out
http/tests/misc/uncacheable-script-repeated.html -> timed out
http/tests/loading/load-javascript-after-many-xhrs.html -> timed out
http/tests/appcache/top-frame-3.html -> timed out
http/tests/appcache/reload.html -> timed out
http/tests/appcache/progress-counter.html -> timed out
http/tests/appcache/progress-counter.html -> timed out
http/tests/appcache/main-resource-hash.html -> timed out
http/tests/appcache/main-resource-hash.html -> timed out