Closed lalexgap closed 1 year ago
The rationale for this change makes sense to me. The testground docs don't really give a clear steer on which API we ought to use, but the tradeoffs seem clear and worth making the change to diagnostics.
Before approving: there are lots of seemingly unrelated changes on this PR. Are they intentional?
The rationale for this change makes sense to me. The testground docs don't really give a clear steer on which API we ought to use, but the tradeoffs seem clear and worth making the change to diagnostics.
Before approving: there are lots of seemingly unrelated changes on this PR. Are they intentional?
Good catch! I think some changes were added via a rebase. I'll clean that up.
Fixes #128
When using results (
runenv.R()
) all the metrics are sent to the database at the end of the run. If the run is long this can be quite a large set of data, which fails to get sent to the database with the errorRequest Entity Too Large
.This switches most of our metrics back to diagnostics (
runenv.D()
) which immediately writes out to the database. The only metrics that are still posted toresults
are theci/nightly
summary metrics. Since these are summary metrics about the whole test run it seemed more appropriate for them to post to results.This does mean that most of the metrics are now excluded from the dashboard link on the tasks page. However the dashboard link is to a bunch of autogenerated graphs that aren't really useful, so I don't think that's a big loss.
TODO: