whats a resource to you in this context?

whatever you're issuing a check on

ok

a global ZedToken is generally not necessary, unless you have concerns about consistency globally but still want to use the cache

does the performance of the check degrade over time if writes are rare, and the token we store doesn't change much?

w/ at_least_as_fresh

if you're using

at_least_as_fresh

, then there is no appreciable difference after a short period of time between that and

minimize_latency

i see

because once all the caches are beyond the timestamp of the ZedToken, it is the same thing

at_least_as_fresh

with a ZedToken just says "make sure the cache has caught up; if not, don't use the cache"

once the cache is past it, it doesn't matter

so if you store a ZedToken and then use it an hour later, all the caches will be well beyond that ZedToken and it becomes

minimize_latency

i see.

so the only reason we would want to cache on write is to close the 5 second gap

correct, and only if you need it

caching on check is redundant and really just moving the problem

by "caching" you mean storing the ZedToken?

yes

then yes

ok

if you care about the 5s gap, then ideally you store the ZedToken solely when you update the content of whatever resource you're checking

and solely for that resource

you don't have to do it that way, but that's the recommended approach

alright. thanks! thats rather helpful

anytime 🙂

hi, in the latest spicedb I see "max costs" default values are percentages. How are they determined in a containerized environment? In my case I have no "limit" but request is set, and I see in the logs these costs change at every deployment. I suppose, with no limits, they're based on the memory available on the system when the process was started, am I correct?

that is correct, it's based on the memory available in the machine. You can also set an actual number if you don't want to go with a percentage.

It suspect there's a memory leak in SpiceDB. My pods are regularly restarting (36 times in 5d18h), with memory going very high pretty quickly. That only started happening since I upgraded to v1.13.0

Hey all, is there any forum or area where schema modelling questions can be discussed or an archive where I can check on previous questions? I have some questions on RBAC cases where I am unsure if SpiceDB is able to support those cases (at least I did not find a solution via the modeling guide or schema reference).

Hi is there a way to reverse a search on expected relations ? given this simple model definition user{ } definition document{ relation owner : user relation reader : user permission read = reader + owner } document:document1#owner@user:user1 document:document2#reader@user:user1 I can then get the document that the user owns: Expected Relation : document:document1#owner: - "[user:user1] is " but not all document that the user is related to ( document1 and document2 ) with somthing like "user:user1#document" or can I ? I could ad a reverse relation but the I would need to make two entries user->document and document->user each time, which seem difficult to maintain.