[Nagiosplug-help] check_ssh: Client says "Read from socket failed: Connection reset by peer"

[Stefan Lasiewski]

Hello everyone,

I am checking most of my Unix systems using the 'check_ssh' check. It seems that check_ssh initiates an SSH session to the client, and then simply drops the connection.

As a result, the following critical error appears in my system logfiles (At /var/adm/messages, /var/log/messages, or /var/log/auth.log).

I am unable to replicate this error by hand every time. It only seems to happen less then 30% of the time.

To replicate, I ran the following command:

$ ./check_ssh -H HOST1 && date
SSH OK - OpenSSH_3.7.1p2 (protocol 1.99)
Wed Dec 10 21:51:12 PST 2006

In the corresponding client logfile, I see the following error:

Dec 10 21:51:12 HOST1 sshd[3641]: [ID 800047 auth.crit] fatal: Read from socket failed: Connection reset by peer
Dec 10 22:11:21 HOST1 sshd[5703]: [ID 800047 auth.crit] fatal: Read from socket failed: Connection reset by peer
Dec 10 22:26:14 HOST1 sshd[7173]: [ID 800047 auth.crit] fatal: Read from socket failed: Connection reset by peer


Is there a way to configure the check_ssh check to behave better? 'auth.crit' is a pretty important log facility, and I don't want to clutter it with noisy messages regarding Nagios SSH checks.

I appreciate any help, thank you!


-= Stefan

--
Stefan Lasiewski | System Administrator | GetActive Software