Custom query fails, no recovery option in UI
I tried a custom query, just ran the default
.take(10) from the current table, didn't make any edits. It errored, but there is no indication of what the error was in the custom query panel or in the logs.
There's also no controls visible to do anything further, the run function button is now missing (see screenshot)
27 Replies
Looks like the missing run function button is a separate bug - if you expand the custom query panel and then collapse, the button disappears below the fold.
Deployment name?
Happening in all envs on a couple of different projects, here's a dev env: curious-lark-383
I have an old version of my app and a newer project that I migrated to a couple months ago to get around the import/export issues with older projects. Both have the issue. But another project created in between those two doesn't have it, and new projects don't have it. So it's something specific to my app.
Do you have Bytes stored in the table you're trying to query?
No
I've tried the default query on a very simple table with a few records, same
Added a test table and convex is automatically adding it to my schema and deploying, that's wild
Seems to be stuck though
If you can figure out what's special about the deployments where the custom query fails that would really help
Trying
Any idea why it's stuck validating? I can't see any logs or anything for that on my end
RE: the root issue, the screenshot up top shows "Error" - there's nothing for that in your logs?
looks like a 500
Where are you deploying from?
I didn't deploy, the validation spinner is in the UI, presumably because I added a table through the UI (for testing)
It added a schema entry for the table I added and started redeploying
Via
npx convex dev?No I'm not running anything
This is all through UI interactions in the convex dashboard
1. added a table through the UI
2. added a document with a string field
3. observed the generated schema in the dashboard showed an entry for the new table
4. observed the new schema being validated message
I went ahead and:
1. deleted the table through the UI, as custom query showed inline errors due to the table not being in the schema
2. added the table into my schema locally and deployed to dev
3. added a document to the table in the UI
4. observed that the "A new schema is being validated after a code push..." continues to show, even after a successful deploy
5. attempted default custom query against the new table and it stsill returns the same 500
Can you share details of the 500 from the Network tab?
I shared the payload above, anything else specifically? The request has an admin key, I can dm you the request body
Can you try to reload your dashboard (for the schema validation being stuck issue)
Did that a bunch
Did again for good measure
Note: this is just a bug report, completely non-blocking for me
Just to make sure I understand -- sounds like there's 2 bug reports here.
(1) being stuck on the "new schema is being validated" spinner in the dashboard after pushing code
(2) custom test query failing for unclear reasons
And both of these happen on
curious-lark-383 (and also some other projects)
For (2), we do see an error on our side but need to do some more investigation to figure out what exactly is happening there.Correct - it is not clear whether these two issues are related as I only happened to notice 1 while troubleshooting 2.
Actually, just checked an unrelated project and it also has issue 1, and I haven't made a change to that project in months. Might just be a UI or state bug.
If you click Show Schema in a project and scroll to the bottom you may see it in your projects too
It looks like it specifically shows on my projects that have schemas defined in code, those that don't do not have this behavior.
Aha yeah I also see (1) in my projects with schema!
hooray for easy repros 🎉
Yep (1) seems to just be an inverted boolean in the dashboard UI code and will be fixed shortly
One other small UI bug from earlier in the thread: https://discord.com/channels/1019350475847499849/1219995332306403389/1219995732510113944
^ I'll be working on that one today too 🙂 thx for reporting these issues
@erquhart when you get a chance can you try the custom query again? We weren't able to reproduce the issue ourselves so curious if you still can or has resolved. We have added some additional logging and instrumentation to help us narrow it down if it happens again.
@presley Still failing, same 500 but it did print an error message in the UI this time:
It also tried to send a payload to sentry, but got a 413 payload too large, in case you were hoping to get insights from that. I copied the payload if you want me to DM it to you. The payload appears to include an array of all DOM objects in the dashboard, hence the payload too large error.
Ok we believe we've identified the issue on our side! Will keep you posted on a fix
I believe this should be fixed now! (let us know if that's not the case)
Works! Thank you 🙂