Folks,
I have two different CouchDB clusters to which I applied the same
changes -I wanted to avoid view and database compactions that were
eating away at their performance.
I applied the same changes to both clusters using the API, and both
clusters returned the same smoosh configuration:
{
"view_channels": "upgrade_dbs,ratio_dbs",
"db_channels": "upgrade_dbs,ratio_dbs"
}
(I dropped the slack criteria.)
However, while one cluster has stopped compacting, the other has not.
While both cluster have the same databases, they run on different
configurations:
- the behaving cluster is a four-instances, CouchDB 3.1.0 cluster
running on Kubernetes (Apache CouchDB Helm chart 3.3.4),
- the misbehaving cluster runs on four separate VMs, each with a
Docker container based on the Bitnami 3.1.0 image.
Could someone explain this behaviour?
Cheers,
Luca Morandini
