OHOS: Robustify the runners #46
Open
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
- Upgrade Dependencies - Have a new function `check_and_inc_retries` that will panic if the retries is too much without success. We will exit the process and crash. If we succeed we set the retries to zero. - Unify processor timeout throughout main loop. Some paths through the main loop had some timeouts while some had none. We now have a unified timeout at the end (before killing offline runners) of 5 seconds. - Read the env variable "RUNNER_SUFFIX" to add a suffix to the runner name, so we do not have to hardcode it. - Introduce a new function `call_github_runner_api` which encapsulates simple github api calls with gh. `spawn_runner` now uses this new functionality for spawning. Previously some - Introduce `kill_offline_runners` which calsl the github api for current runners, takes runners that are offline and have a name starting with `dresden-hos`. Then it removes these runners. This happens at the end and 5 seconds after we started the runners. The runners need some time to start and connect to github properly. The timeout seems ok tbut we can probably increase it. Sorry for the massive changes but I think the codebase is so small that it is allowed :D Signed-off-by: Narfinger <[email protected]>
jschwe
reviewed
Jul 25, 2025
Member
jschwe
left a comment
jschwe
left a comment
There was a problem hiding this comment.
Did you also have a look at the github runner run.sh and check if we can determine if the runner spawned successfully by wrapping run.sh?
|
|
||
| thread::sleep(Duration::from_secs(5)); | ||
| // Check if some still running images are listed as offline from github api point of view | ||
| if let Err(e) = kill_offline_runners(&servo_ci_scope) { |
Member
There was a problem hiding this comment.
I don't think we want to do this all the time, only when we detected a problem. github API calls are rate-limited, and the limit is not even too high.
Signed-off-by: Narfinger <[email protected]>
delan
force-pushed
the
main
branch
2 times, most recently
from
87b9cb4 to
2f00564
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This involves a couple of changes:
check_and_inc_retriesthat will panic if theretries is too much without success. We will exit the process and
crash. If we succeed we set the retries to zero.
main loop had some timeouts while some had none. We now have a unified
timeout at the end (before killing offline runners) of 5 seconds.
name, so we do not have to hardcode it.
call_github_runner_apiwhich encapsulatessimple github api calls with gh.
spawn_runnernow uses this newfunctionality for spawning. Previously some
kill_offline_runnerswhich calsl the github api forcurrent runners, takes runners that are offline and have a name
starting with
dresden-hos. Then it removes these runners. This happensat the end and 5 seconds after we started the runners. The runners need
some time to start and connect to github properly. The timeout seems ok
but we can probably increase it.
Testing: Tested on runner CI2. It succesfully removed old offline runners on github and started new ones.
I plan to let CI2 run this new code for a bit to see if there are any regressions without changing CI1.
Hopefully this fixes: #44
Sorry for the massive changes but I think the codebase is so small that
it is allowed :D