And some notes from Claude to the
And some notes from Claude to the containers team
12 Replies
Im sure the cloudflare container's team will appreciate the nice dose of AI slop, can't believe they missed it
You have more chance of winning the powerball lottery than this problem being an issue related to POSTs within Cloudflare's control, your container is crashing and restarting for whatever reason - the logs already tell you this, you've implemented a workaround rather than getting to the root of the problem
We're working on some stuff to surface errors better (exit code, reason, etc) but yeah, the likely culprit is the user process died and so our request got cut-off.
Stackdriver back in the day had a nice feature where you could search for a log entry say "Appliation started" and then see the context around it, i.e. it wouldn't filter to just that but would scroll that into view so you could see the page of stdout/stderr before the crash or before restart (~same)
I added something for the next dash release where the recent errors dropdowns of the dash have a "view logs" that takes you to the logs page for that specific placement & filtered to 5 mins either side of the error
nice, that sounds good
I do think it'd be neat if our logs are in there too so you don't have to rely on the recent errors part though
i.e today if your process gets killed for out of memory, your logs will just end - you have to check the separate Recent Errors display to see the VMStopped error
you mean like aggregate the logs into single timeline so VMStopped etc
ah
yep
definitely be very helpful to see infra and app logs timelined together
I think that'd mean you also get them in any Logpush setup and so you can send them to... whatever
last week a problem, this week not so much if the commitment to rollout all features to all users is carried out in a reasonable timeline
i.e. the no more enterprise feature blog post
Workers Trace Events is available to anyone on Workers Paid
No Enterprise needed for that dataset
will take a look thanks, im probably* a little out of date on whats been made available over last year
All the others that were already around like HTTP Requests are still Enterprise (well, maybe not in the near future since that blog)
but Workers Trace Events (your Worker logs and your Containers logs since we join in there) is anyone on Workers Paid