Commands
# spicedb
k
Hi folks, can the default value of
--datastore-gc-interval
[option](https://authzed.com/docs/spicedb/concepts/commands) (3m) have impact on the latency of spicedb grpc calls by any chance? we're read heavy and mainly using
Consistency_FullyConsistent
mode for most of our operations.
For context, we're seeing and debugging very random spikes across read endpoints leveraging spicedb listing grpc.
v
using full consistency is not going to scale. You are bypassing all caches.
all the heavy lifting is done by the database, not SpiceDB
so you have to scale your DB accordingly
we've addressed issues with GC in the past, so I'm moderately confident those are addressed now
it definitely used to cause spikes in the past because of unindexed queries
k
Gotcha, we started with full concurrency but planning to slowly optimize and move to low latency mode. Thank you for confirming that this isn't related to issues with GC!
6 Views