Description
Currently we are running the CI jobs on ASF infrastructure as simple unix applications, and are sharing the machines with different projects (and different instances of our jobs).
This results in quite a few problems:
- Clobbering other jobs/getting clobbered by buggy cleanup efforts in scripts
- Causing/Suffering from resource starvation
- Conflicts in shared maven local repositories (f.e. HBase recompile)
The obvious solution to (some) of these problems is utilising Jenkins's docker support.
One problem with that approach is that we lose the efficiency of a persistent local maven repo as a cache. We may try to modify the maven setup to use that, or we can accept the performance hit, considering that our jobs are very long, so the additional time/traffic generated is not a deal breaker.
Attachments
Issue Links
- is depended upon by
-
PHOENIX-6111 Jenkins jobs are unable to create new native thread
-
- Resolved
-
- is related to
-
PHOENIX-5032 add Apache Yetus to Phoenix
-
- Resolved
-
- links to
