you re using the operator or

Question

mgagliardo · Answer

yes installed the operator and the basic CRD in the example ``` apiVersion authzed com v1alpha1 kind SpiceDBCluster metadata namespace spicedb play name dev spec config image ghcr io authzed spicedb v1 13 0 datastoreEngine memory secretName dev spicedb config apiVersion v1 kind Secret metadata namespace spicedb play name dev spicedb config stringData preshared key token ```

Joey · Answer

< ecordell> might have some insight it appears to be a dispatching issue

mgagliardo · Answer

Once I got it running and port forwarded I could use zed CLI to write read schema and write permissions

mgagliardo · Answer

but the permissions check command fails

ecordell · Answer

do you see any errors in the logs of the spicedb pods

mgagliardo · Answer

``` level error requestID a9726bbbbd2bf3a880dee394428f2350 error rpc error code = Unavailable desc = last connection error connection error desc = transport Error while dialing dial tcp 10 30 59 76 50053 connect connection refused time 2022 10 06T20 07 32Z message received unexpected error level warn grpc component server grpc method CheckPermission grpc method type unary grpc service authzed api v1 PermissionsService peer address 127 0 0 1 57468 protocol grpc requestID a9726bbbbd2bf3a880dee394428f2350 grpc start time 2022 10 06T20 07 32Z grpc code Unavailable grpc error rpc error code = Unavailable desc = last connection error connection error desc = transport Error while dialing dial tcp 10 30 59 76 50053 connect connection refused grpc time ms 0 459 time 2022 10 06T20 07 32Z message finished call ```

ecordell · Answer

ah dispatch should be disabled for `memory` datastores I ll work on a fix in the meantime you can remove the `SPICEDB DISPATCH UPSTREAM ADDR` env var from the deployment or you can use a non memory datastore

mgagliardo · Answer

alright I ll give that a try I was just starting with memory to test it out

mgagliardo · Answer

is there a way to remove that env var via the `SpiceDBCluster` CRD or would I need to do it manually

ecordell · Answer

it would need to be manual yeah the operator won t overwrite it unless you make other changes to the SpiceDBCluster object

ecordell · Answer

thanks for bringing this up definitely a regression

mgagliardo · Answer

separate question but I figure you may be good to ask is there an official helm chart that we could use as well We are pretty helm heavy on our end and like the ability to keep track clean up deployments via releases I wasn t sure if the operator could give us that same functionality

ecordell · Answer

there s not yet but we re looking at it https github com authzed spicedb operator issues 76

ecordell · Answer

if you want to throw your vote behind helm

ecordell · Answer

ah an easier option for you

ecordell · Answer

``` spec config datastoreEngine memory dispatchClusterEnabled true image ghcr io authzed spicedb v1 13 0 ``` if you add `dispatchClusterEnabled true ` it will enable the dispatch server even with the memory datastore and the pod will connect dispatch to itself

mgagliardo · Answer

Awesome threw a comment in that issue you linked