You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The DRM tests have network dependencies that result in races. Independent of if the tests are accessed locally or at http://w3c-test.org/ they need to make requests to the license server, so there is always some variable lag due to this. However, even in relatively straight forward tests, such as drm-events this appears to result in rare intermittent fails, even when running locally (for my case a license request must round tripped from New Zealand).
I've been looking at adjusting some of the test timeouts, but it's a bit of a whack a mole. I'm raising this issue so further discussion can be had around how to best handle this.
The text was updated successfully, but these errors were encountered:
I suggest all .drm. tests have their timeout extended across the board. Then we don't need to keep playing whack-a-mole and up the timeouts as we trip over intermittent failures caused by latency with the DRM license server.
I guess the question is: Are they reasons to limit these tests to 10 second execution limits? The only downside might be a bit longer execution time on valid timeout results. @cpearce; Do you have an alternate time in mind (e.g. 20s)?
Yes, that would be my main concern. Looking at the latest results, we don't have many timeouts, so I think this should be rare and thus okay. (We want tests to fail rather than timeout if possible. See #3938. In some cases, such as depending on an event, it is not possible.)
The DRM tests have network dependencies that result in races. Independent of if the tests are accessed locally or at http://w3c-test.org/ they need to make requests to the license server, so there is always some variable lag due to this. However, even in relatively straight forward tests, such as drm-events this appears to result in rare intermittent fails, even when running locally (for my case a license request must round tripped from New Zealand).
I've been looking at adjusting some of the test timeouts, but it's a bit of a whack a mole. I'm raising this issue so further discussion can be had around how to best handle this.
The text was updated successfully, but these errors were encountered: