though i don't know enough about grpc-go to have an answer for you

in java it does generate them and you can reference them off the class

👋 I've just made a

schema write

where I removed 3 permissions. I was expecting this to be instant, but it took ~3min, during which i observed a huge increased load on my (Postgres) database due to this query:

SELECT namespace, object_id, relation, userset_namespace, userset_object_id, userset_relation FROM relation_tuple WHERE created_transaction <= $1 AND (deleted_transaction = $2 OR deleted_transaction > $3) AND namespace = $4 AND relation = $5 LIMIT ?

How comes removing a permission requires a DB query? How comes it is so expensive? My DB CPU jumped from ~10% to 50-60% for 3 minutes: this was my staging environment, I'm concerned about deploying this to production :/

👋 I ve just made a

schema write

where

I've tried to upgrade to v1.10.0, but my SpiceDB server doesn't seem to get healthy 🥲 SpiceDB starts:

json
{"level":"warn","prefix":"grpc","time":"2022-08-03T08:58:35Z","message":"grpc server serving plaintext"}
{"level":"warn","addr":":8443","prefix":"http","time":"2022-08-03T08:58:35Z","message":"http server serving plaintext"}
{"level":"warn","addr":":9090","prefix":"metrics","time":"2022-08-03T08:58:35Z","message":"http server serving plaintext"}
{"level":"warn","addr":":8080","prefix":"dashboard","time":"2022-08-03T08:58:35Z","message":"http server serving plaintext"}

But the health probe fails:

/ # /opt/grpc_health_probe -v -addr=localhost:50051
parsed options:
> addr=localhost:50051 conn_timeout=1s rpc_timeout=1s
> tls=false
> spiffe=false
establishing connection
connection established (took 904.622µs)
service unhealthy (responded with "NOT_SERVING")

I'm not getting any log (with LOG_LEVEL=warn) telling me anything's wrong, how can I investigate further?

Try log level trace

health checking did change a bit in 1.9.0 so that it asserts the health of the datastore

Picking up the question of Google Workspace sync from a few weeks ago, Google has essentially confirmed that one cannot have a SCIM integration unless it is offered as a marketplace app (note that it is unclear if that is sufficient either), so I'm going to write up some stupid poll/sleep loop for now. However, it isn't obvious to me how to deal with multiple changes, esp with nested groups. As far as I've seen, there's no bulk relationship write? Is there a way to avoid clients potentially seeing half-changes, in particular when group-in-group memberships change?

Hi guys, I've run into a situation where the LookupResources call fails, but I can't really figure out what the problem is. It occurs in our integration tests, where we set up some data ( create entityA), then delete permissions to alter this entity, and then try to edit entityA. This leads to the LookupResources call. This is where it goes wrong, does anybody know what exaclty is going on here (full file is also attached)?

{"level":"error","module":"pgx","args":[],"err":"timeout: context canceled","pid":62,"sql":"begin isolation level serializable read only","time":0.0711,"time":"2022-08-04T06:41:56Z","message":"Exec"}

The

WriteRelationships

call is a batch call: it takes zero or more updates to apply in a single call

looks like a timeout; is this against

serve-testing

?

@here SpiceDB v1.11.0 was just released! Highlights include: - REST gateway can now be enabled for

serve-testing

to allow REST API tests - Reachability now uses batching to significantly improve the performance of LookupResources on wider datasets - Check requests can now have their trace returned for debugging purposes (changes in Zed will go in shortly to enable this from the CLI) - New concurrency limits to prevent too many goroutines from being spawned - CRDB feature detection for the Watch API no longer admin permissions, enabling use against CRDB serverless - WriteSchema no longer checks permissions for data, improving update performance where only permissions are removed https://github.com/authzed/spicedb/releases/tag/v1.11.0

@Arash ^ do you want to give 1.11.0 a try against CRDB serverless?

Hey Gurus, do you have playground URL handy for the github example discussed in video ,

It would be really great if you can share it

its in the description of the video

oh nice, thanks for share friend !

@Joey @Jake I have a local 3 node cockroach cluster running in insecure mode. Started spicedb with that as the datasource. I have run the spicedb migrate . However when i run the zed command , i run into issues

Please help.

You need to make sure your zed context knows it’s insecure

yes, added that too.

zed context set first-dev-context :50051 "somerandomkeyhere" --insecure

Your preshared key must be wrong now

oops, sorry -this was what i needed to share

Can you try putting the schema in a file and passing the file name?

same. Just checking - the preshared key is the same as the one passed to spicedb serve grpc_shared_key param - correct?

yes, it's the same, and if that were the problem you would see the error about invalid token

are you using the exact same datastore parameters for

spicedb serve

as you did for

spicedb migrate

?