-
Notifications
You must be signed in to change notification settings - Fork 175
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
v1.0.3: panic: interface conversion: interface {} is string, not float64 #433
Comments
I was able to grab another stack trace, this one seemingly more complete, for a similar panic that happened a little bit after the one that opened up this issue:
I looked around to see if there was any recorded interaction with the socket |
Thanks for sharing this panic. This is an interesting one because it means the data in the |
Very interesting. It's puzzling to us because we've had literally billions of queries go through that Trickster and then suddenly panic after panic with this type error out of nowhere. We still haven't been able to figure out what the query string was, and now that we're back on 1.0.3 we haven't had any of them since last week. ¯\_(ツ)_/¯ I'll roll out beta 4 when it's ready and see what we catch in the logs! |
@WesleyDavid closing this one as you indicated you've been using v1.1.5 without problems |
Trickster Version:
1.0.3 buildInfo: 2020-04-16T14:37:36+0000 41fbf794f4457d1d6f3453ab480612d5eb10ae92 goVersion: go1.13.8 goArch: amd64
Cache kind: in-memory.
Backend: InfluxDB 1.7.
Host OS: CoreOS / Linux
Runtime: Docker
There's a decent amount of traffic coming through the cluster 60 - 90 queries per second. We're mirroring some customer queries that would normally go straight to a time series backend. That mirrored traffic is getting pushed through a Trickster. As of last night, we noticed Trickster had crashed out, which is very uncharacteristic. So far we've never seen a crash. The stack trace is rather short:
Prior to that, things appear to be normal. We've got debug logging turned on and requests were coming in and returning with status code=200. These were the last few lines just before the panic above (I removed query responses with code=200 because it shows customer data):
On our end, the infrastructure side of things, we haven't (knowingly) changed anything. We've been letting it run for days to gather statistics for us to analyze. However since this is a mirror of customer traffic, there's any number of arbitrary queries that could be coming through. Current working theory is that some new query that is otherwise absorbed by InfluxDB (either by responding with a
200
or just gracefully erroring out and returning a400
or500
response) is perhaps tripping a bug in Trickster.Some Context
As a result of this experience, we decided to do a quick test of the latest beta build, 1.1 beta 3, which resulted in seemingly unrelated crashes within minutes of the process being up. That experience resulted in this issue, which may or may not be related, but the context and timing seemed relevant to note.
The text was updated successfully, but these errors were encountered: