chore: fix tests and CI flakiness #16806
Conversation
gkalpak
force-pushed
the
chore-try-to-fix-CI
branch
3 times, most recently
from
3ae49db to
097aa69
Compare
gkalpak
force-pushed
the
chore-try-to-fix-CI
branch
5 times, most recently
from
2df3547 to
d68189b
Compare
gkalpak
force-pushed
the
chore-try-to-fix-CI
branch
18 times, most recently
from
a8eae83 to
eae7e59
Compare
This includes a Karma fix that affects CI flakiness. Based on angular/angular#27735.
gkalpak
force-pushed
the
chore-try-to-fix-CI
branch
from
eae7e59 to
b96d334
Compare
|
The latest Safari on Saucelabs (v12) is very unstable and disconnecting all the time, so I removed it from the browser list. |
| @@ -189,6 +184,9 @@ module.exports = function(config, specificOptions) { | |||
| config.sauceLabs.tunnelIdentifier = process.env.TRAVIS_JOB_NUMBER; | |||
| config.sauceLabs.recordScreenshots = true; | |||
|
|
|||
| // Try 'websocket' for a faster transmission first. Fallback to 'polling' if necessary. | |||
| config.transports = ['websocket', 'polling']; | |||
|
|
|||
There was a problem hiding this comment.
Is this not the default?
There was a problem hiding this comment.
According to the docs, the default is ['polling', 'websocket'] and it is not clear if ordering makes a difference, so I thought it wouldn't hurt to match angular/angular#27634.
But, yeah, this might be redundant.
There was a problem hiding this comment.
It seems that socket.io will
By default, a long-polling connection is established first, then upgraded to “better” transports (like WebSocket).
So if I understand it, ['polling', 'websocket'] would connect with polling initially and then try to connect via a websocket. Seems strange but I guess that polling comes up quicker for short-lived connections?
| CONNECT_STDERR="$LOGS_DIR/sauce-connect.stderr" | ||
| # We don't want to create a log file because sauceconnect always logs in verbose mode. This seems | ||
| # to be overwhelming Travis and causing flakes when we are cat-ing the log in "print_logs.sh" | ||
| CONNECT_LOG="/dev/null" |
There was a problem hiding this comment.
TBH, I don't know if it was actually affecting flakiness. I was just porting bits that looked useful from Angular PRs. (This came from angular/angular#27657.)
| @@ -7,5 +7,5 @@ for FILE in $LOG_FILES; do | |||
| echo "================================================================================" | |||
| echo " $FILE" | |||
| echo "================================================================================" | |||
| cat $FILE | |||
| cat $FILE || true | |||
There was a problem hiding this comment.
Can this cat fail?
There was a problem hiding this comment.
If there are no log files, then for FILE in $LOG_FILES uses $LOGS_DIR/* literally, which does not exist and thus cat fails.
gkalpak
force-pushed
the
chore-try-to-fix-CI
branch
2 times, most recently
from
75997bc to
b96d334
Compare
What kind of change does this PR introduce? (Bug fix, feature, docs update, ...)
CI-/Test-related chores.
What is the current behavior? (You can also link to an open issue here)
CI is broken and extra flaky.
What is the new behavior (if this is a feature change)?
CI is neither broken nor flaky.
Does this PR introduce a breaking change?
No.
Please check if the PR fulfills these requirements
Fix/Feature: Docs have been added/updatedFix/Feature: Tests have been added; existing tests pass