This troubleshooting guide describes what to do if you experience slow Scylla Manager behavior or when connections to Scylla nodes over SSH are timing out.
This might affect users of the Scylla Manager when determining REST API status of the managed clusters. Scylla Manager Client might report certain nodes as being down even if they are accessible.
Scylla Manager manages the Scylla nodes over the HTTP API. Communication between Scylla Manager server and Scylla nodes is encrypted by tunneling HTTP traffic over an SSH connection.
Establishing an SSH tunnel requires that Scylla nodes have a running sshd server and optionally the sshd server can be configured to do reverse DNS for resolving client IPs. When resolving takes a long time or it stalls for some reason then connections and interaction with the sshd server can be impaired.
There are two options for solving this.
One option is to improve your DNS setup on the Scylla node by changing to a better DNS resolver. It is recommended to use static nameserver IPs from Cloudflare or Google.
Another option is to disable reverse DNS usage for the sshd server. This can be done by navigating to the sshd configuration file (located in /etc/ssh/sshd_config on Unix based systems) and setting the UseDNS configuration option to off.