Skip to content

Fix builds and advance Linux runner to ubuntu-20.04#65

Merged
philrz merged 2 commits intomasterfrom
current-actions-runners
Aug 29, 2022
Merged

Fix builds and advance Linux runner to ubuntu-20.04#65
philrz merged 2 commits intomasterfrom
current-actions-runners

Conversation

@philrz
Copy link
Contributor

@philrz philrz commented Aug 29, 2022

As noted in brimdata/brimcap#259 (comment), while the changes in #64 allowed the build automation to run successfully and produce an artifact, the result actually ended up being broken. Specifically, the Suricata Update binary produced by pyinstaller complained of being unable to find a module.

$ ./suricataupdater 
Traceback (most recent call last):
  File "suricata-update", line 32, in <module>
ModuleNotFoundError: No module named 'suricata'
[2060] Failed to execute script 'suricata-update' due to unhandled exception!

I could repro this breakage even with the Ubuntu artifact, and I'd not touched that build automation at all in #64, so this indicated something else had changed on us in the time since we last built.

I hacked at this and found that I could make a working binary if I just rolled back to the version of pyinstaller as of when we last successfully built. Through further binary searching I found the first pyinstaller version that breaks is 4.6, so here I've pinned to the version right before that. I don't see anything in their 4.6 changelog entry that sticks out as an explanation for the change, but I'm inclined to go with what works for now and look into making it work with current pyinstaller as need arises.

While I was under the hood, I remembered how GitHub is set to retire some of the older versions of the Actions Runners, so I took a shot at advancing those. For Linux, the move from ubuntu-18.04 to ubuntu-20.04 only needed a couple of library tweaks, so I've included that here. An attempt to make the macOS move from macos-10.15 to macos-11 was not so simple. In the interest of divide & conquer, rather than get into that here I'd like to merge this PR so I can get working artifacts that include the fix to brimdata/brimcap#259 and then look into the Mac build stuff at a later date.

To test, I created a draft Suricata artifact https://github.com/brimdata/build-suricata/releases/tag/v5.0.3-brim3, then pointed at that from a test Brimcap suricata-v5.0.3-brim3 branch that ran successfully through CI here and also a test Brim suricata-v5.0.3-brim3 branch that pointed at that Brimcap commit, and that ran successfully through CI here.

@philrz philrz requested review from mattnibs and nwt August 29, 2022 18:22
@philrz philrz self-assigned this Aug 29, 2022
Copy link
Member

@nwt nwt left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No need to do anything about this here, but for future PRs, it's preferable to confine orthogonal changes like these to separate PRs.

@philrz philrz merged commit 5c55786 into master Aug 29, 2022
@philrz philrz deleted the current-actions-runners branch August 29, 2022 19:06
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants