Plugin goes down when used with Grafana Docker Container.

[ghost]

Hi!
So, I observed something weird when working with this plugin and using the grafana docker container.

I started my grafana docker instance using this command,

docker run -d --name=grafana -p 3000:3000 grafana/grafana-enterprise

Grafana started up without any issues. I was able to access it on port 3000 via the browser.

Then, I went to the plugin's section via the gear icon on bottom left of the homepage. Screenshot shown below

After this, I installed this plugin.

Now, when try to create a panel, I get this error: "An error occurred in the plugin". This happened before I can type anything.

I see these logs from the grafana container,

logger=installer.fs t=2022-11-11T09:24:56.7883905Z level=info msg="Downloaded and extracted grafadruid-druid-datasource v1.4.0 zip successfully to /var/lib/grafana/plugins/grafadruid-druid-datasource"

logger=plugin.loader t=2022-11-11T09:24:57.0169687Z level=info msg="Plugin registered" pluginID=grafadruid-druid-datasource

logger=plugin.grafadruid-druid-datasource t=2022-11-11T09:25:25.693593Z level=info msg="DRUID EXECUTE QUERY" _________________________GRAFANAQUERY___________________________="map[Interval:1.5e+10 JSON:map[datasource:map[type:grafadruid-druid-datasource uid:9-6KULD4k] datasourceId:1 intervalMs:15000 maxDataPoints:1488 refId:A] MaxDataPoints:1488 QueryType: RefID:A TimeRange:map[From:2022-11-11T03:25:25.648Z To:2022-11-11T09:25:25.648Z]]"

logger=context userId=1 orgId=1 uname=admin t=2022-11-11T09:25:25.6984401Z level=error msg="Internal server error" error="[plugin.downstreamError] failed to query data: Failed to query data: rpc error: code = Unavailable desc = error reading from server: EOF" remote_addr=172.17.0.1 traceID=

logger=context userId=1 orgId=1 uname=admin t=2022-11-11T09:25:25.6985415Z level=error msg="Request Completed" method=POST path=/api/ds/query status=500 remote_addr=172.17.0.1 time_ms=12 duration=12.3085ms size=116 referer="http://localhost:3000/dashboard/new?editPanel=2&orgId=1" handler=/api/ds/query

After this, any attempts to add a query result in continuous errors

logs says this,

logger=context userId=1 orgId=1 uname=admin t=2022-11-11T09:28:44.0246852Z level=error msg="Internal server error" error="[plugin.unavailable] plugin unavailable" remote_addr=172.17.0.1 traceID=

logger=context userId=1 orgId=1 uname=admin t=2022-11-11T09:28:44.0248356Z level=error msg="Request Completed" method=POST path=/api/ds/query status=500 remote_addr=172.17.0.1 time_ms=2 duration=2.1185ms size=98 referer="http://localhost:3000/dashboard/new?editPanel=2&orgId=1" handler=/api/ds/query

I am assuming that an empty query was issued by grafana to the plugin. This caused the plugin to crash and not recover after that. However, I am not sure why this happens only on the grafana container. These exact steps seem to work fine when I run grafana locally.

An interesting thing is the difference in the log statement, the docker plugin has this _________________________GRAFANAQUERY___________________________ whereas, when run locally, there is a space between GRAFANA and QUERY. I can also see that the log being printed here shows a space between GRAFANA and QUERY

Could it be that the versions getting installed on the container and locally are different? I verified these and found them to be the same.

The same behavior was also observed when I hosted this docker instance on AWS EKS.

I will be happy to fix this myself with a few pointers.

[ghost]

@jbguerraz I'll be happy to fix this with some pointers.

[saketbairoliya2]

@MohammadGhazanfar - I was facing similar issue but installing it using cli fixed the issue. Also, you can configure the druid credentials directly in docker/grafana/provisioning/datasources/druid.yml. Let me know if this helps.

[ghost]

Thanks @saketbairoliya2 !
I placed a config file in provisioning/datasources and that seems to work around this!

[ghost]

Keeping this open because it still tracks an issue with the plugin.

[alvaro-ponce]

Any news?
Facing the same problem after upgrading from grafana 8.5.15 to 9.3.2

[jbguerraz]

Hello @alvaro-ponce , @MohammadGhazanfar,

How is the datasource configured ? does the connection test pass there ?

[ghost]

Hey @jbguerraz!
Configured the data source from the Grafana UI. Yeah, the connection test passes.

[ghost]

@alvaro-ponce you can work around this by provisioning the datasource using a file. See @saketbairoliya2 's above.

[jbguerraz]

I indeed reproduced the issue @MohammadGhazanfar

Interesting is that if you go to explore at first, then run a query, it works.
Or if you restart the docker container, then it also kinda works (plugin crash but get restarted).

The bug on the plugin side is here: https://github.com/grafadruid/druid-grafana/blob/master/pkg/druid.go#L399
somehow Grafana sends a query without Builder (despite https://github.com/grafadruid/druid-grafana/blob/master/src/QueryEditor.tsx#L16 ) and the code is not ready for that: we better make sure there is map and not a nil value before using it.

Maybe here https://github.com/grafadruid/druid-grafana/blob/master/pkg/druid.go#L392 , with something like:

if err != nil || q.Builder == nil || q.Settings == nil

Feel free to fix it @MohammadGhazanfar

[ghost]

Thanks @jbguerraz ! I will look into this and submit a PR!

[ghost]

@jbguerraz Created a PR here for this: #138

[jbguerraz]

Thank you @MohammadGhazanfar ! Closing this issue since #138 is merged now 👍