Abnormal usage
We got a huge amount of read bandwidth today on our dev environment, 240mb nothing abnormal on any other graph to indicate high usage or lots of queries. Ran empty imports a few more times etc, We are usually at like 20-30.. tempted to say its a bug?
6 Replies
Do you use a lot of HTTP actions?
ya we have a lot of failing webhooks from testing but wouldnt that reflect in a higher than normal function count?
Not crazy, on clerk Successful Attempts
444
Failed Attempts
894 in 6 hours of dev
Could you open a support ticket via email? There has been a change to HTTP bandwidth reporting (should be more accurate now)
Yeah, even when i was solo programming without much webhook (like 30) it was like 40mb
no response yet, not exactly a paying customer but have barely passed the 1gb. I did find the bandwidth option under functions breakdown by projects which accounts for it. I'm still not 100% sure why it was so bad those two days. But ive never seen the bandwidth by function report until after this. We get a ton of re-renders when we re-init with the npx convex import. I think these paginated queries expand with new db entries + re-rendering which i started to mitigate. Now i should redo the init process. 🙃 We still got 2/3rds more queries to make before mvp
Sorry for the delay here, we'll look into it to confirm. I think this is because we've corrected the bandwidth accounting, for HTTP functions. I thought you should email so we're sure to look into it. I believe the new accounting is correct, but we'll confirm Monday.
We managed to do it again, this time a bit different, a infinite loop caused 7000 events to be added to our db with now much decreased payloads etc. Just this time it was another order of magnitude. My event monitoring page managed to eat 4.5gb of db bandwidth. It never threw a error. But I assume it had 1000's of re-renders of it growing to 1000's of rows even with useStablePagination
I cant find anything about how to control the size of paginated queries if they grow as a result of new database entries
I'm also assuming doing this effectively causes all the events table data to be read when it otherwise wouldnt be.
.table("events")
.order("desc")
.filter((q) => q.neq(q.field("type"), "events"))
.paginate(args.paginationOpts)