Schema syntax error....

hey guys sorry for disturbing you. but why did i get one prefix of

blog/

very appreciate it if anyone can help on this

no idea about this situation, so for this case, should that be something wrong with my spiceDB instance?

okay got it. when we use the spice db, we have to give one prefix for all definition

but still have no idea about why the permission check is wrong 😦

Check uses the cache; you likely need to give a zedtoken from your recent write call or wait 10s for the cache to expire

See https://docs.authzed.com/guides/first-app#checking-permissions

oh i see

thank you so much

BTW can the client be the singleton? will the connection of singleton client be broken if it lasts for one long time? Thank you !

Hi! Chances are we haven't generated the newest version of the client. Let me have a look.

what client language are you using? It's gRPC so it's HTTP2, I'd be surprised if the generated client wasn't thread-safe, but I'd need to check

@Joey - Would you mind sharing the link to your Kubecon presentation ? The one on Twitter doesn’t seem to work

are you referring to kubecon 2022? What video are you looking for?

The one presented in cloudnative Rejekts in October 2022

one sec

looks like it isn't posted yet

keep an eye on https://twitter.com/aamit01singh/status/1584747510488530944

Hi All, I am new to SpiceDB. I have a requirement to install Spice DB instances in GCP through cloud run. Please share the details/step by step document if we have any.. Thanks in advance

it was posted, but somehow removed?

we don't have any documentation for Cloud Run. Another community member made it work in AWS Fargate, so maybe this could be of help? https://github.com/authzed/examples/pull/5

I'm having serious issues trying to upgrade to 1.14.1 on Postgres (manual, no-downtime upgrade process). I've done the

add-xid-constraints

migration, running

write-both-read-new

, and I'm observing much higher response time. Initially the response time shot through the roof (>1s P95), it seemed to be a pod CPU issue rather than a database issue, so I mitigated by doubling the # of replicas. It brought back the response time within usable range, but it's still much higher than it was on 1.13.0: my P50 is 3x higher (~15ms) and my P95 is 2x higher (130ms). I'm not seeing any particular pressure on the database, and the CPU usage of my pods is way higher than it used to be (>0.5, used to be ~0.1), so I assume something is burning CPU that wasn't before. I have this problem with

write-both-read-new

but also

write-both-read-old

. I have traces but I'm not getting any insight from them, maybe you do? I'm concerned about whether my cluster will be able to handle the load on Monday: I'm already breaching my SLOs with these response times on a Sunday where we have low traffic 😰

Hey, just wondering if it's possible to add metadata/tags to relations in SpiceDB. I'm hoping to use this so that i can tag a set of "static" relations that are synced from a config file. My process needs to create these relations from the config file and delete any relations that were removed from the config file. If spicedb supports metadata/tags then I can use that to find relations that need to be cleaned up. Is this a thing?

currently, no. it is recommended to use other relationships to track such data if necessary. there is an issue for this: https://github.com/authzed/spicedb/issues/966

hmm okay, so i should write the required relation and then store an additional relation to mark it as one of my static ones. given any relation in my system can come from a config file it's not practical to add an extra relation for each relation in my schema; so i guess i need to use a subject for this specific use-case.

"mything:1#can_greet@user:1" and then "mything:1#can_greet@user:from-config-file"

or perhaps encode the relation as a b64 string and have a dedicated resource for tracking "static_relations:BASE64ENCODED#is_static@?"