Fixing OUT_OF_DISK errors
When you provision a Typesense Cloud cluster, we give you 5X disk space, where X is the amount of RAM you pick.
The amount of disk used by Typesense is completely dependent on the amount of data you've stored in Typesense.
Once the amount of data you've sent into Typesense exceeds available disk space in your cluster, you'll start seeing OUT_OF_DISK errors.
To fix this issue, you can go into Cluster Configuration > Modify pick a higher memory configuration and schedule an upgrade. This will in turn give you a larger disk.
We frequently see disk space filling up on clusters because of stale writes that get accumulated due to short timeouts for write API calls in the client libraries.
Writes in Typesense are synchronous. So if you have a short client-side timeout when instantiating the client libraries, the client might terminate the connection before the write is fully complete.
To remedy this, you want to configure a client-side timeout as high as say 15 minutes, so the write API calls never timeout mid-flight.
To clear stale data from disk, you can trigger a disk compaction, followed by a snapshot. If you find yourself doing this often even after increasing the client-side timeout, please reach out to us to with your Cluster ID to increase the frequency of the automated disk compaction (that runs every 7 days by default).