fix(concourse): Fix caching of concourse build events #525
Conversation
| @@ -124,8 +124,13 @@ private JobDelta jobDelta(ConcourseProperties.Host host, Job job) { | |||
| List<GenericBuild> genericBuilds = | |||
| builds.stream() | |||
| .map(build -> concourseService.getGenericBuild(jobPath, build, false)) | |||
| .filter(b -> !cache.getEventPosted(host, job, cursor, b.getNumber())) | |||
There was a problem hiding this comment.
We still want to run cache.setLastPollCycleTimestamp() for jobs without uncommitted successful builds, which filtering here prevents.
There was a problem hiding this comment.
Ah okay. I should be able to add a cache.setLastPollCycleTimestamp() in the if right after this then right?
I believe we need the filter + return null there when there are no uncommitted successful builds or else the JobPollingDelta list will always be the # of concourse jobs, correct?
Just want to make sure I'm understanding the problem.
Nvrmind that seems to make me get stuck in a loop where it continually is pushing and caching everything over and over again.. Will take a look and figure out correct way to fix this.
Figured it out.
donnyyung
force-pushed
the
fix_concourse_fastforward
branch
2 times, most recently
from
c1f9197
to
1c77330
Compare
Builds should be added to cache correctly now and the admin fastforward command should work spinnaker/spinnaker#5047
donnyyung
force-pushed
the
fix_concourse_fastforward
branch
from
1c77330
to
4b522c4
Compare
|
@spinnakerbot cherry-pick 1.16 |
Builds should be added to cache correctly now and the admin fastforward command should work spinnaker/spinnaker#5047
|
Cherry pick successful: #530 |
Builds should be added to cache correctly now and the admin fastforward command should work spinnaker/spinnaker#5047
…spinnaker#530) Builds should be added to cache correctly now and the admin fastforward command should work spinnaker/spinnaker#5047
Builds should be added to cache correctly now and the admin fastforward command should work
It solves spinnaker/spinnaker#5047
I ran into this problem because we had over 1000 builds on our concourse instance, so we were running into
cache exceeds upper threshold (1000).Upon trying to fix this, I realized the builds were never being compared to what was in the cache until the commitDelta function, which unfortunately we don't get to unless the threshold is less than 1000.
Note: This is my first PR to a spinnaker project and my first time coding in java so open to any and all suggestions!