https://signoz.io logo
#support
Title
# support
g

Godspower Unuane

08/29/2023, 5:37 PM
is it just me or is the selfhosted version unstable
v

Viljar Võidula

08/29/2023, 5:38 PM
I've been using it for 1.5 months , been stable so far , no complaints, only thing to watch out is configuration , depending on setup might need to change some things
g

Godspower Unuane

08/29/2023, 5:40 PM
just keep getting this after every deployment
v

Viljar Võidula

08/29/2023, 5:41 PM
ah its probably the query service dying , k8s deployment?
g

Godspower Unuane

08/29/2023, 5:41 PM
no docker-compose
v

Viljar Võidula

08/29/2023, 5:43 PM
recommend going over the logs - , might have more info, i had similar issues with way too large sampling on my apps and tracing enabled. Ran a single instance clickhouse on some cheap EC2 instance with gp2 storage
i turned my sampling rate to 0.2 and increased storage to get more iops and was a lot better after
g

Godspower Unuane

08/29/2023, 5:43 PM
i havent even connected it to anything yet
v

Viljar Võidula

08/29/2023, 5:44 PM
oh 😞 then i dont know 😞
on k8s the deployment takes quite a bit of time tho , for example when running on minikube
it takes like 15 minutes for me to switch all the pods
u

Ulisses Silva

08/29/2023, 7:26 PM
how did you changed the sampling rate, @Viljar Võidula? Using the settings from this page ? https://github.com/open-telemetry/opentelemetry-collector-contrib/tree/main/processor/probabilisticsamplerprocessor
s

Srikanth Chekuri

08/30/2023, 12:15 PM
Sampling can happen at the collector or at the agent level (less sophisticated). @Viljar Võidula is probably referring to head based mode.
v

Viljar Võidula

08/30/2023, 12:16 PM
sorry for late reply . i have self instrumented app and i modified sampling rate from the app code. Didnt know about collectors work at that time yet .