-
Notifications
You must be signed in to change notification settings - Fork 13.3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[FLINK-21346][build system] Adding timeout to all tasks #15546
Conversation
Thanks a lot for your contribution to the Apache Flink project. I'm the @flinkbot. I help the community Automated ChecksLast check on commit 6213fbc (Fri May 28 09:08:20 UTC 2021) Warnings:
Mention the bot in a comment to re-run the automated checks. Review Progress
Please see the Pull Request Review Guide for a full explanation of the review process. The Bot is tracking the review progress through labels. Labels are applied according to the order of the review items. For consensus, approval by a Flink committer of PMC member is required Bot commandsThe @flinkbot bot supports the following commands:
|
The Azure failure is expected to show the layout of uploaded files. Once the change is reviewed I will remove the |
Regarding manual compression: I think it is safe to remove the manual compression as it is done automatically by azure:
with manual compression:
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thank you for picking that PR up!
Downloaded misc and there is a superfluous file in there.
Downloaded e2e and all looks good.
I like that we got rid of the double compression. Makes it easier to check.
@@ -81,7 +81,7 @@ jobs: | |||
condition: and(succeeded(), not(eq(variables['MODE'], 'e2e'))) | |||
pool: ${{parameters.test_pool_definition}} | |||
container: ${{parameters.container}} | |||
timeoutInMinutes: 240 | |||
timeoutInMinutes: 60 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
For the final PR this should be set to 2h or so (lower than the 4h we used before).
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Why should we lower it? We did not make the tests run faster...
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The test didn't need 4h and they run quicker into a potential timeout this way. Just check the build times on recent builds (btw 4h is the default).
Which superfluous file do you have in mind? Is it |
displayName: Upload Logs | ||
inputs: | ||
targetPath: $(ARTIFACT_DIR) | ||
artifact: logs-${{parameters.stage_name}}-e2e |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I added the stage name to the artifact name to avoid duplicate file issues in the nightly CI profiles. It seems that the stage name is done again based on timestamps, which has the risk of artifacts not being uploaded, because they exist already (IIRC it has happened that files had exactly the same timestamp (since builds are triggered at the same time)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I reintroduced the stage name.
Besides these two comments, I didn't spot anything in the changes. |
As for the compression, let's try to remove it for now as it makes it a bit easier to work with the files. It lets you also download a single file from a build. If we hit a problem with a resource limitation we can easily reintroduce the compression. |
…stacktraces. The test/e2e task is killed before timeout and artifacts are published in the regular way. If killed jps traces and watchdog output are additionally attached to mvn.logs.
a9291f1
to
5141fb3
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM % squash fixups. Thank you for your contributions.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think this change breaks the local e2e test setup on my machine (MacOS). The problem is that $FLINK_DIR/logs
does not exist.
@@ -32,30 +32,24 @@ if [ -z "$FLINK_DIR" ] ; then | |||
exit 1 | |||
fi | |||
|
|||
if [ -z "$FLINK_LOG_DIR" ] ; then | |||
export FLINK_LOG_DIR="$FLINK_DIR/logs" |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think this change breaks the e2e setup on my machine (MacOS) because the directory does not exist.
@@ -46,6 +46,10 @@ if [ -z "$FLINK_DIR" ] ; then | |||
exit 1 | |||
fi | |||
|
|||
if [ -z "$FLINK_LOG_DIR" ] ; then | |||
export FLINK_LOG_DIR="$FLINK_DIR/logs" |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Same here.
@@ -26,6 +26,10 @@ if [[ -z $FLINK_DIR ]]; then | |||
exit 1 | |||
fi | |||
|
|||
if [ -z "$FLINK_LOG_DIR" ] ; then | |||
export FLINK_LOG_DIR="$FLINK_DIR/logs" |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
And here.
Can you please fix it? Dawid is MIA afaik. Or maybe we should have a |
Maybe, I will just fix the problem by setting |
What is the purpose of the change
I took over #14834
When a test gets stuck, it's usually killed after 10 min of inactivity. #13260 excluded all tests that still show log activity from this timeout to allow test runs that take longer than 10 min.
However, some stuck tests are not deadlocked but rather caught in some kind of livelock. These tests now timeout with AZP timeout (4h) without any logs. This commit also adds a timeout to all build tasks in the same way as e2e tests are timeboxed. On reaching the timeout, the task logs are uploaded together with a stack trace.
Verifying this change
Manually verified.
Does this pull request potentially affect one of the following parts:
@Public(Evolving)
: (yes / no)Documentation