Closed hackartisan closed 1 month ago
It's true, they aren't, and the monitor is not alerting, even though the watcher correctly indicates that it's down:
irb(main):025:0> FileWatcherStatus.new.check!
/opt/figgy/releases/20240514232304/app/checks/file_watcher_status.rb:11:in `check!': FileWatcherStatus::FileWatcherStatusCheckError (FileWatcherStatus::FileWatcherStatusCheckError)
Update: I had a typo, this endpoint is correct, it's just not reporting the failure for some reason even though calling the class directly does give the error as expected (see above)
the provider configured into datadog isn't hitting a valid endpoint
https://github.com/pulibrary/princeton_ansible/pull/4804/files
Not sure what the endpoint should be.
Okay it looks like the way the filewatcher gem works was changed; maybe this is the cause of that expected hash got string error. see differences between the previous readme and current readme:
https://github.com/filewatcher/filewatcher/blob/v1.1.1/README.md
watch.do used to yield filename, event
but now it yields changes
which itself yields filename, event
.
The update happened a month or so ago https://github.com/pulibrary/figgy/commit/5c8e9bbcd9ff8610e368e2deeceed3dc65da8d42#diff-89cade48462044ee1b672dc5f4c3ec250fbd29effcd8932096a23c1283c6731fR386
the timing with the honeybadger error starting just yesterday afternoon doesn't make much sense, but still this seems to match the error.
rolling back the filewatcher gem stopped the error but you still have to touch all the files that are in there so they're not old.
we did find . -type f -name "*.pdf" -exec touch {} +
the files are getting processed now and I let Tracy know; closing
Also, the monitor started working again when we deployed figgy.
User report via lsupport
Sudden priority justification
staff work is blocked delivering documents to users