When there are erroneous lines in known_hosts, Mina will write a large amount of duplicate data into the known_hosts file.

[zhniu]

When the known_hosts file gets corrupted due to certain reasons (such as concurrent SSH writes), Such as:

Mina will throw an exception while parsing the known_hosts file.
Even if there are valid public keys in the known_hosts file, they won't be read out.
This causes the system to write a new public key into the local known_hosts file every time it connects to a server. After running for a long time, this will result in the known_hosts file occupying a very large amount of disk space.

The code:

mina-sshd/sshd-common/src/main/java/org/apache/sshd/client/config/hosts/KnownHostEntry.java

Line 206 in c10dbcd

} catch (RuntimeException | Error e) { // TODO consider consulting a user callback

[tomaswolf]

Yes, the current implementation throws an exception on the first error. Which is one of the reasons why JGit uses its own known_hosts parser.

Thanks for reporting this; I had completely forgotten about this.