ServerAliveInterval - rundeck is NOT observing this?

[Jennifer Fountain]

I think it was resolved here - https://github.com/rundeck/rundeck/issues/897 - but still doesn't working as I would expect it.  

Here is my ssh config:

[rundeck@rundeck1 .ssh]$ cat config 

Host *

    ServerAliveInterval 300

    ServerAliveCountMax 2

If there is a setting in rundeck that does observe this setting, I can't find it anywhere in documentation.

Help!

Thanks

[Greg Schueler]

You can configure SSH config options for the built-in SSH plugins in Rundeck in this way: http://rundeck.org/docs/plugins-user-guide/ssh-plugins.html#specifying-ssh-config-options

however, I have not seen confirmation that the "ServerAliveInterval" setting actually has any effect with the underlying Jsch library that we use.

--
You received this message because you are subscribed to the Google Groups "rundeck-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rundeck-discu...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/rundeck-discuss/2bbe46bb-b6da-4c32-8a83-8ee97d49167b%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Jennifer Fountain]

Thanks Great.  What do you recommend for jobs that run for more than 30 minutes and the firewall disconnects the connection?

[Mmin Hhuber]

@greg: tried with project.properties. however the corresponding tcpdump didn't give me any "push" entries. Here are the project.properties entries for the keepalive (for testing purposes I have used an interval of 5 seconds, so no need to wait too long):
----- SNIP -----
project.ssh-config-ServerAliveInterval=5
project.ssh-config-ServerAliveCountMax=3
service.NodeExecutor.default.provider=jsch-ssh
----- SNAP -----

Hence I switched over to the "classical" approach using script-exec with ssh options in the project.properties file:
----- SNIP -----
plugin.script-exec.default.command=ssh -oServerAliveInterval\=60 -oServerAliveCountMax\=3 -oStrictHostKeyChecking\=no -oLogLevel\=Error -tt ${node.username}@${node.hostname} ${exec.command}
service.NodeExecutor.default.provider=script-exec
----- SNAP -----
and these connection.properties entry gives me every 60 seconds a push output while using tcpdump.

Any idea why the jsch-ssh approach doesn't work?


@jennifer: the ServerAliveInterval value may be different, depending on the other firewall ttl values. We use for our systems 60 seconds. However 600 seconds is also a good value as many firewalls have ttl's as of 1 hour. Just ensure that the ServerAliveInterval is smaller than the ttl on the other devices configured.

hope this helps.