no perf run for rustc PR #93089, #94061

[klensy]

Yet another one. There no (visible) perf run results for PR rust-lang/rust#93089. Queue is empty, but no run for that pr in queue.

[Mark-Simulacrum]

As always, the perf run results are present - https://perf.rust-lang.org/compare.html?start=df368ae457c54fb95d3e64f9986a5f171a6370f0&end=17dfae79bbc3dabe1427073086acf7f7bd45148c&stat=instructions%3Au -- just no summary comment is posted. Investigating that now.

[Mark-Simulacrum]

[2022-01-25T12:51:53Z DEBUG site::request_handlers::next_commit] next_commit: 17dfae79bbc3dabe1427073086acf7f7bd45148c (missing: Master { pr: 93089, parent_sha: "df368ae457c54fb95d3e64f9986a5f171a6370f0", is_try_parent: false }) was logged, which should mean we ran queue_pr -- but I don't see a row in the table from the query:

insert into pull_request_build (pr, complete, requested, include, exclude, runs) VALUES ($1, false, CURRENT_TIMESTAMP, $2, $3, $4) ON CONFLICT DO NOTHING

The table has a couple unique constraints:

   Column   |           Type           | Collation | Nullable | Default | Storage  | Stats target | Description
------------+--------------------------+-----------+----------+---------+----------+--------------+-------------
 bors_sha   | text                     |           |          |         | extended |              |
 pr         | integer                  |           | not null |         | plain    |              |
 parent_sha | text                     |           |          |         | extended |              |
 complete   | boolean                  |           |          |         | plain    |              |
 requested  | timestamp with time zone |           |          |         | plain    |              |
 include    | text                     |           |          |         | extended |              |
 exclude    | text                     |           |          |         | extended |              |
 runs       | integer                  |           |          |         | plain    |              |
Indexes:
    "pull_request_build_bors_sha_key" UNIQUE CONSTRAINT, btree (bors_sha)
    "pull_request_build_pr_idx" UNIQUE, btree (pr) WHERE complete = false

But I'm fairly confident they should not have impeded this row being inserted, so I'm confused why we don't have a row for this query -- maybe the on conflict do nothing is causing trouble? It seems like it should be fine...

[Mark-Simulacrum]

Manually executing the two queries and hitting onpush seems to trigger the comment as before. @rylev / @nnethercote, do you perhaps have some idea for how to continue debugging this failure mode? My best guess right now is to add an ad-hoc select after inserting to try and confirm we inserted, but that feels... terrible and rather odd.

insert into pull_request_build (pr, complete, requested) values (93089, false, CURRENT_TIMESTAMP);
update pull_request_build set bors_sha = '17dfae79bbc3dabe1427073086acf7f7bd45148c', parent_sha = 'df368ae457c54fb95d3e64f9986a5f171a6370f0' where pr = 93089 and bors_sha is null;

[nnethercote]

My secret shame is that I managed to skip out on the supposedly mandatory databases class at university and so I know almost nothing about this stuff. Sorry :/

[rylev]

@Mark-Simulacrum perhaps we could remove the ON CONFLICT DO NOTHING to see in a conflict is occurring (which would then cause a panic).

[Mark-Simulacrum]

I'm pretty sure it's necessary -- the intent is that if there is already a row (e.g., we're queueing twice) that's OK. I guess we can avoid unwrap() and instead log::error! to avoid a panic (which would cause the end point to not return, seems bad).

It does seem like the right next step I guess, though I don't really understand how there can be a conflict with (what seems like) nothing, since there's no rows for the PR in the table.

[klensy]

One more, rust-lang/rust#94061

[Mark-Simulacrum]

Looks like same as before -- not present in pull_request_build, but we did benchmark it.

I am seeing the next_commit logged -- which looks like it should have produced the expected entry in pull_request_build:

[2022-02-16T22:11:18Z DEBUG site::request_handlers::next_commit] next_commit: 75d9a0ae210dcd078b3985e3550b59064e6603bc (missing: Master { pr: 94061, parent_sha: "f90b06d7fb00470177516c5881bcec0432cb4f24", is_try_parent: false })

I guess now that #1166 has been merged we'll try again on the next missed one.