Heikki Linnakangas
Heikki Linnakangas
Rebased again. https://github.com/neondatabase/neon/pull/2189 has now been resolved, so unless some new blockers arise for this PR, this is ready to be committed.
Happened again: https://app.circleci.com/pipelines/github/zenithdb/zenith/1746/workflows/999057a9-bf5a-4d3e-8613-584716a2bf6e/jobs/11086
Got another failure with the same error: https://github.com/neondatabase/neon/runs/7250242872?check_suite_focus=true ``` 2022-07-08 11:20:44.809 INFO [test_wal_acceptor_async.py:338] BackgroundCompute 0 query failed: Multiple exceptions: [Errno 111] Connect call failed ('127.0.0.1', 16153), [Errno 99] Cannot assign...
And a few more here: https://app.circleci.com/pipelines/github/neondatabase/neon/7822/workflows/71c87bd2-5707-40d0-aae1-4fc7993d09fe/jobs/80303, see [tmp/test_output/test_unavailability/repo/pgdatadirs/tenants/57391c9e525f441bb18801c0dedaaf7d/test_safekeepers_unavailability_pg_node/pg.log](https://output.circle-artifacts.com/output/job/981a199f-6526-40cc-a26f-d093ebad3949/artifacts/0/tmp/test_output/test_unavailability/repo/pgdatadirs/tenants/57391c9e525f441bb18801c0dedaaf7d/test_safekeepers_unavailability_pg_node/pg.log) That was from the CI run for the same commit as above, but I don't think there was anything in the PR...
Another one, from a docs-only PR: https://app.circleci.com/pipelines/github/neondatabase/neon/8150/workflows/ef2c3e04-1585-40c3-802b-7ebe7611c673/jobs/83441/steps
> Another one, from a docs-only PR: https://app.circleci.com/pipelines/github/neondatabase/neon/8150/workflows/ef2c3e04-1585-40c3-802b-7ebe7611c673/jobs/83441/steps Never mind, I this was different. I was looking at the postgres log, and it contained errors like this: ``` 2022-07-18 10:49:03.595...
@lubennikovaav, have you thought how we would build v14 and v15 side-by-side? I'm thinking something like this
Here's what it will look like when we add v15 to it: https://github.com/neondatabase/neon/commit/ead121be1d92bbe9aa85b651ba50afebe71fd93b. Lots of stuff remains to do, of course, like building postgres_ffi for both versions, refactoring code to...
This was obsoleted by PR #2337, closing.