[wdspec] Reducing default timeouts for implicit waits, navigation and script execution
Categories
(Remote Protocol :: Marionette, task, P1)
Tracking
(firefox113 fixed)
| Tracking | Status | |
|---|---|---|
| firefox113 | --- | fixed |
People
(Reporter: whimboo, Assigned: whimboo)
References
(Blocks 1 open bug, Regressed 1 open bug)
Details
(Whiteboard: [webdriver:m6])
Attachments
(1 obsolete file)
Given that all the navigation commands in wdspec are done against the local wptserve instance I feel that we should lower the default navigation timeout and not use the default value of 300s as specified by the WebDriver spec.
IMHO 30s should be more than enough and we might even have shorter ones. Tests that require a longer timeout could override it instead.
Lets get it discussed today.
| Assignee | ||
Comment 1•4 years ago
|
||
Lets reduce it as much as possible and test it on try. We could even check with shorter than 30s.
| Assignee | ||
Updated•4 years ago
|
| Assignee | ||
Comment 2•3 years ago
|
||
Note that by default wdspec tests have a default timeout of 25s for the whole test file! That means having 30s as mentioned above would not help at all and the full file would run into a timeout again.
As discussed we could go with 5-10s by default for opt builds and everything that requests pages via wptserve or local. We should apply the appropriate multiplier for special builds. Lets get some try jobs running to verify the correct behavior.
| Assignee | ||
Comment 3•3 years ago
|
||
When removing the triage keyword I missed to set the backlog whiteboard entry. We should keep this on our radar in case of a higher timeout rate.
| Assignee | ||
Updated•3 years ago
|
| Assignee | ||
Comment 4•3 years ago
|
||
I think that we should not blindly set 5-10s but should make it dependent on the type of build. As such it would be good to have the RunInfo available (bug 1814128).
| Assignee | ||
Comment 5•3 years ago
|
||
Basically this should not only be set for the page_load timeout but also for script and implicit (which is 0 anyway). So this will reduce the risk of running into test timeouts due to long script evaluation as well.
For a test I pushed a try build for page_load set to 3s and script set to 1s:
https://treeherder.mozilla.org/jobs?repo=try&revision=0608e05ce06ce0697a5ae2c801354cf8cc9979fc
| Assignee | ||
Comment 6•3 years ago
|
||
Timeouts are based on the timeout multiplier as determinted
by the wptrunner to allow longer timeout for slow running builds.
Updated•3 years ago
|
Updated•3 years ago
|
| Assignee | ||
Updated•3 years ago
|
Updated•3 years ago
|
| Assignee | ||
Comment 7•3 years ago
|
||
The patch has been merged upstream and will by synced with the next wpt downstream sync on bug 1822836.
| Assignee | ||
Comment 8•3 years ago
|
||
Fixed by the downstream sync on bug 1822836.
Description
•