Examples of rendering:
And failure: https://build.bitcoinabc.org/viewLog.html?buildTypeId=BitcoinABC_BitcoinAbcStaging&buildId=91243&guest=1
Don't keep the original message in case of a failure
Deduplicate the status message
- Actually write to the log BEFORE publishing the artifacts...
- Override the build status message in case of a failure
flash => flush
Sync to make sure the build logs are complete before aborting
Rebase on top of D6817 to see failure messages
Talked to @jasonbcox and the server will be upgraded to more memory
The stress test is still running full capacity on all other tests, and when running with TSAN we are especially looking for deadlocks rather than performance issues.
The point still stands, this patch was a solution I can leverage by myself to work around the issue now but I will look at if/how we can increase the machines capacity.
Pass by reference
Thu, Jul 2
Move the test argument decalaration where it belongs
Wed, Jul 1
The above comment will trigger a build failure, don't take it into account.
Tue, Jun 30
The bot build ran from master, there is a bug there
Can you rename the UBSAN exception accordingly ? See D6784
Mon, Jun 29
Relayout the usage comment
If you rebase on top of D6763 you only have a single occurrence of the seeder target remaining in contrib/teamcity/builds/build-osx.sh
Please keep the targets renaming for another diff as there are some places where this needs to be updated as well.
Sat, Jun 27
More context: the bash traces are echoing a list of the valid targets, as it checks for validity when a list is passed to build_with_cmake. This list has no line break and causes the overflow.
Fri, Jun 26
Setting the job pool for custom targets is only possible with cmake >= 3.15 and we enforce cmake >= 3.13 (debian 10 default): https://gitlab.kitware.com/cmake/cmake/-/merge_requests/3308.
Note that due to the wrapper, USES_TERMINAL will not dump the test to the console unless there is a failure. It's still not equivalent to a dedicated pool but close enough.