Fall back to install_and_cache_pkgs.sh if restore_pkg.sh fails

[torsknod2]

• Changed post-cache-action.sh to fall back to install_and_cache_pkgs.sh if restore_pkgs.sh fails
• Ensure restore_pkgs.sh also fails if substituted commands fail

[awalsh128]

One concern is unexpected behavior and silent fails that extend the time of execution and doesn't leverage the action. What scenarios do you think this would be helpful?

[torsknod2]

@awalsh128 Whyever I had the issue that I had a cache hit, but the content restore did not work.
This is why I came to the idea to have a fallback for such situations.

After having that I noticed that the restore exits with 0 even with the error during restoring from the cache when files are missing.
Because of that I added the trap.

Unfortunately I was never able to really root-cause it, because I found no way to download the files from the cache to my local machine when workflows failed.

[awalsh128]

Do you have possible some old logs I could look at, or maybe the list of packages and version you were at? Also how long ago was it. The cache package has seen an upgrade.

[torsknod2]

@awalsh128 I am on a business trip now until end of the week, so I have only the phone to access my personal stuff.
The runs with the actions are used in here https://github.com/torsknod2/MDSpanTest/actions/workflows/ci.yml .
The most recent ones use my fork, but a bit older runs use yours.
Best is to filter for a branch starting with number 2.