[ https://issues.apache.org/jira/browse/CASSANDRA-18665?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17771583#comment-17771583 ]
Josh McKenzie commented on CASSANDRA-18665: ------------------------------------------- Looks like you get pre-github-PR style JIRA comments from me on this one. :D - Agree w/Maxim; I think we should drop support for anything not on the maintained list here: [https://cassandra.apache.org/_/download.html]. There's a non-trivial burden associated with us carrying these scripts, support, and artifacts along. - Consider removing duplication of supported versions between {{cassandra-artifacts.sh}} and {{{}cassandra-dtest-pytest.sh{}}}; maintenance error-prone. - Same for regx_version in {{cassandra-test.sh}} (i.e. having a shared .sh they all source w/a function that returns the regx would simplify) - Consider {{mkdir -p}} instead of {{mkdir}} in execute call in {{prepare_release.sh}} to be a bit more fault tolerant. Or nuking and re-creating. /shrug - Line 288: wildcard {{cp build/cassandra*}} may end up catching things we don't want, either now or in the future. Is that the right scope for the glob? h3. *{{cassandra_job_dsl_seed.groovy}}* - Is {{dtest-upgrade-upgrade}} a typo on line 74? Only occurrance I could find in either this repo or C* proper. Maybe it's supposed to be {{{}dtest-upgrade-large{}}}? - Why 600 for the timeout on artifact building? Does that include queue time as well or just execution? i.e. if the latter... 10 hours is a long time. :) - In our buildSteps in {{{}matrixPostBuildScript{}}}, it might be nice to get some kind of echo of a) how much data we pruned from tmp and /tmp, and b) how much data we're still using. Something like: \{{find $1 -type f -exec du -ah {} + | sort -rh | head -n 20}} (or whatever # suits your fancy) has been helpful for some of my testing to try and find what big consumers are. - Kind of nitty, but is there a reason we \{{cloneOption { shallow(false) }}}? Seems we'd save some space by doing shallow of branch only. - Why are we testing on python3.6? (i.e. we're defaulting to 3.7 in {{{}docker/run-tests.sh{}}}?) - Have a diff between {{HEAD}} and {{c2aa282}} camping out around line 823 - Is there any way to get away from the significant redundancy of specifying the clean commands on each job? - nit: indentation on Branch Pipelines header comment h3. *{{.build/docker/ubuntu2004_test.docker}}* - So the \{1..11}seq for ccm creation - what's the failure rate and reason look like there? Why 11 attempts? - If we initialize the ccm repos w/cassandra builds at docker image build time, we're getting a pretty baked in version of things that needs to be updated each time we cut a new release right? Why do this at docker image compile time rather than runtime, and/or just rely on {{{}cassandra-dtest/upgrade_tests/upgrade_manifes.py{}}}'s list of latest versions to correctly pull down and build cassandra instances then? - Along with that, we pre-cache C* versions on python 3.7 env right? Not on 3.6 or 3.11 which we pull down on the image? h3. *{{upgrade_base.py}}* - What's the story with the df and mount debugging on cluster install_dir? > Update jenkins groovy dsl to use in-tree scripts > ------------------------------------------------ > > Key: CASSANDRA-18665 > URL: https://issues.apache.org/jira/browse/CASSANDRA-18665 > Project: Cassandra > Issue Type: Task > Components: CI > Reporter: Michael Semb Wever > Assignee: Michael Semb Wever > Priority: Normal > Fix For: 5.0.x, 5.x > > -- This message was sent by Atlassian Jira (v8.20.10#820010) --------------------------------------------------------------------- To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org