Repository navigation
Blocks E2E: Fix db connection issues - #51658
Conversation
75ab5aa to
94ba9ac
Compare
Test using WordPress PlaygroundThe changes in this pull request can be previewed and tested using a WordPress Playground instance. Test this pull request with WordPress Playground. Note that this URL is valid for 30 days from when this comment was last updated. You can update it by closing/reopening the PR or pushing a new commit. |
|
Size Change: 0 B Total Size: 2.38 MB |
c2cf57c to
6d3b05b
Compare
6e926ca to
3fc11a7
Compare
This reverts commit cdf10d232d2b62a2b529ef6ae8f7d7772aa4d893.
This reverts commit a76dbd375e70e14bb6c4d31fc284707587fa7049.
This reverts commit 9ae1d3b9608cfe6f95358dcc81f537f991bac4a9.
4b7849a to
411d45d
Compare
|
This issue is keeping me awake. After a bunch of failed attempts (see commit history), I thought I had it by disposing of the request cache – I figured we're running out of resources because we're making a lot of requests. Unfortunately, it failed again after a couple of reruns (debugging this one takes forever). I obviously tried replicating the issue locally, without luck. It always passes, even when running entire shards. Anyway, it's time to ask for some help. Here's what I've learned so far:
Example of a failed run. It usually happens within the shard no. 4, which runs the following suites in the following order:
The (currently) failing test is this one: From the failed test's trace.zip, the network requests are as follows:
The part that is completely out of order is the request to activate the Note You can download the trace.zip and open it via As I write this, I’m in the process of cleaning up the numerous Product Collection E2E tests, particularly the product-collection.block_theme.spec.js file, and splitting them up further. These tests currently occupy the CI for the longest time, which is problematic. Many of them involve nested loops and are heavily parameterized, making them challenging to debug and follow. I’m working on simplifying things in the hope that I might either spot an issue or inadvertently fix it. If anyone has a better approach, I’m all ears! 🙏 The following issues might be related:
cc: @gigitux @adimoldovan @adamziel @kevin940726 @Mamaduka @talldan I'm pinging some core folks as it might be related to the |
|
I've not seen a similar error in the Gutenberg repo, but I started noticing tests marked as flaky due to the |
|
Thanks for working on this and for your great exploration, @WunderBart! 🙇 Did we try to use this patch and see if the issue disappears? In this way, we can ensure that the issue isn't related to WordPress. |
@gigitux yep, tried it in 8a59383. The DB kept disconnecting, unfortunately :( |
|
I think we need to narrow down the issue. Is this a problem with Unfortunately, I don't have enough expertise with cc @noahtallen |
|
It could be a dumb thought but I noticed that the last time that we updated If the 8a59383 didn't work, it means that could be some regression related to WordPress/WooCommerce?
💯 I agree! |
|
Hi @lanej0, @louwie17, @adimoldovan, Apart from reviewing the code changes, please make sure to review the testing instructions and verify that relevant tests (E2E, Unit, Integration, etc.) have been added or updated as needed. You can follow this guide to find out what good testing instructions should look like: |
|
Thanks for the feedback so far, folks. I was trying to narrow it down but I just could not repro it locally yet. Will keep trying. In the meantime, after the cleanup, I restarted the workflow several times, and the usual suspects didn't fail. Let's hope it stays like that. Marking this one as ready for review, let's see how it goes! 🙌 |
lanej0
left a comment
There was a problem hiding this comment.
Frustrating that it's one of those problems we can't nail down. No concerns from me, I merged trunk just to trigger another CI run and tests passed -- so 👍🏼 -- we'll just keep an eye out post-merge.

What?
Context: p1727172851985989/1726604625.727459-slack-C03CPM3UXDJ
Closes: #51101 (This test was more flaky than others because of this issue so it was skipped. This PR unskips it.)
Trying to fix an intermittent issue with E2E tests in CI, where a DB gets disconnected mid-suite, failing the job:
See #51658 (comment) for more details.
How to test