Hi team i was deploying signoz v0.18.1 throught k8...
# support
r
Hi team i was deploying signoz v0.18.1 throught k8 , all service were running except otel-collector and otel-collector metrics . i got logs as Error: cannot build pipelines: failed to create "clickhousetraces" exporter, in pipeline "traces": error connecting to primary db: code: 170, message: Requested cluster 'cluster' not found 2023/04/19 101530 application run finished with error: cannot build pipelines: failed to create "clickhousetraces" exporter, in pipeline "traces": error connecting to primary db: code: 170, message: Requested cluster 'cluster' not found . as i checked in clickhouse DB for clusters , there is some cluster list , how to get the cluster name whick otel-collector is looking
s
Are you using external ClickHouse with SigNoz helm?
s
in case of external clickhouse, please set your clustername as cluster
I faced similar kind of issue.
r
i am running all the service in k8 cluster as individual pod with this otel docker image signoz/signoz-otel-collector:0.66.7
clustername in which file i should configure , in clickhouse ? can you share me the file name which i need to change plz ?
s
Let me share the repo link @Ragulraj
https://github.com/hassanbsee2071/clickhouse Just clone this repo and do docker-compose up -d, This will create 6 nodes based clickhouse server on a single node. I am still testing thats why I am using single node. Cluster name is set in macros.xml and remote_server.xml
s
What did you use to deploy to Kubernetes? If you use SigNoz helm charts, it automatically sets it up. If not, you need to make sure there is a cluster named
cluster
in CickHouse.
r
No i am not using signoz helm charts , this is k8 deployment file i prepared for clickhouse
f not, you need to make sure there is a cluster named
cluster
in CickHouse. this in which file i need to configure ?
in this below files i couldnt get clusters config
s
If you hand-roll your own manifests, it will not be possible for me to help since I don’t have any idea what your config will look like.
r
based on the docker-compose.yaml from signoz docs , i created a k8 deployment for each service
my k8 deployment will be almost similar like docker-compose.yaml
in clickhouse DB there is no cluster named "cluster"
Hi @Srikanth Chekuri i was trying now to setup signoz with SigNoz helm charts provided in docs
few pods were running some were in init , seems there is issue in pvc connecting with pv
the pvc are in pending state
s
@Prashant Shahi can help with the helm deployment
p
It case of external ClickHouse, there are some pre-requisites for external clickhouse: • UDF script and histogram quantile binary needed in clickhouse: custom-function.xml and histogramQuantile binary • must have at least one zookeeper instance • distributed cluster named
cluster
set up required: clickhouse-cluster.xml • user credentaials passed must have enough privileges to create/manage databases in CH • secure connection not supported between clickhouse and signoz components yet
@Ragulraj It is likely that you are missing zookeeper as well as including a distributed cluster named
cluster
in clickhouse config.
r
Hi @Prashant Shahi as we are facing many difficulties in external clickhouse , i started a new installation of signoz through signoz helm
s
@Ragulraj I am using external clickhouse, I also shared my repo to install, I am not facing any issue with external clickhouse
r
the repo you shared is for docker-compose , i am using k8
s
docker-compose to install clickhouse server. My signoz is running on AWS EKS
clickhouse server is currently running on ec2 and all other signoz resources are deployed on eks cluster
r
Ok acutally i am moving with signoz helm now , for easy understand
@Prashant Shahi @Syed Muhammad Hassan here pods were still in pending
s
These are my pods, @Ragulraj check your pv and pvc
r
my pvc is still in pending
s
I think this is not related to signoz issue, check how you are provisioning pv. In my case we are using dynamic storage class for provisioning PV.
r
do i need to create any pv manually
s
Try to create PV and bind it pvc manually
Yes if you do not have storage class
for dynamic provisioning
r
Ok, what type of storage class i should give in my PV
if you have any sample can you share me please ?
s
where your k8s cluster is running
r
it is running in OCI server
check this documentation
r
Ok i will check with this , thank you so much
and you created individual pv for the pvc created from signoz helm am i right ?
s
no, my pv got automatically created when I defing volumeclaimtemplate
r
where you defined volume claim template , pv created automatically means by helm ?