I think we concluded that the answer is 'no'

@AndyRussG this is fixed and deployed, right?

One last bit to review, so we record the variant: https://gerrit.wikimedia.org/r/379079

This works!

For CiviCRM code:

PHP client library: https://github.com/Jimdo/prometheus_client_php
Metric types: https://prometheus.io/docs/concepts/metric_types/

Looks like this is a bit more complicated than the SmashPig directories, since we're asking for arbitrary dirs under 'variant'. Will do this with code deploys under the fundraising/REL1_27 branch for now.

Looks like two different windows:

• xxxx:19:09 Donor starts a CC donation in one window
• xxxx:20:16 Donor gets redirected to GlobalCollect
• xxxx:25:08 Donor starts a non-CC donation (in another window?)
• xxxx:25:12 Donor comes back from GlobalCollect

Donor ends up at the ResultSwitcher page with a session that's not CC. We fail it and send a bad message to payments-init.

The 'variant' code is deployed to prod, just needs a change to fundraising_code_update: T175871

The other possibility is to add an setting like ignore_overtime to the process-control job description, which would suppress failmail when the previous job is still running at the time of the next run.

@Eileenmcnaughton , I saw your IRC comments about pitfalls of restarting CSV processing.

This turned out to require upgrading our minFraud SDK and basically rewriting the filter, so I'm upping the point value. On the plus side, the new SDK allows us to actually test with fake HTTP responses.

Patch for review is just the diagnostic part of this ticket.