vanadium-www-tutorials-js-web is occasionally timing out #520
Comments
|
Adding @jakline as he's the next build cop |
|
This has failed three out of the last four times. |
|
I intend to add retry logic within the test itself so that we can avoid this problem. That helps for the short-term. The longer term plan is to debug this using the vanadium-browser-test-web build. (See vanadium-archive/browser#84) I have a wild theory: if any UI test was run recently, it becomes more likely for subsequent tests to succeed. While the browser-test-web was active yesterday, this test succeeded pretty much the whole time. Since it was disabled at ~9:30 today, it's started failing more. This test is scheduled to run 7 minutes after the browser test... All of this implies some sort of remote cache is involved, which is why we'll be looking at the identity server next. |
|
Hi Alex, It will be good to have a fix. It failed seven times in a row this morning. #398-404 When do you plan to add your retry logic? I ask because we are expected to try to push for fixes to flaky tests during our build cop rotation. Arup stated yesterday that Asim was going to add some logging to identityd in an attempt to rule it in or out. Asim might have some relevant info today. |
|
I'm writing the CL that will add the retry logic, which should get in later today. Re: identityd. Yup, I'll take a look later at those logs. |
|
Btw, Asim checked in his identityd change last night. John and Jatin can sit down and do the daily release after lunch, and then it will be in production. |
The caveat page doesn't always show up. vanadium/issues#516 This has affected other tests vanadium/issues#520 vanadium-archive/browser#84 In order to improve test stability, the test will retry after a 60s delay. Refreshing the page will very likely allow the test to continue and check what it really ought to be checking. The actual bug with wspr/identityd will have to be looked at separately. Log lines were added to assist in debugging this problem. - screenshots are taken of the first caveat error - IP addresses are printed at the start of any web test - if the test fails, all console output is printed out (did not print this upon first caveat error, might want to do that) MultiPart: 1/3 Change-Id: I3b8cbaa6867a67bde6174e6f51ad164dff8123b9
|
This test has been quite stable recently, so we'll continue tracking the identityd issue separately. |
Alex, can we increase the timeout on this one? It has timed out twice today, and thereafter run successfully when I retry it. Both successes and failures are on node11, so I don't think it's a problem with the machine.
The text was updated successfully, but these errors were encountered: