Doug,

What about something like monit to make sure ssh is up and running and restart 
if it crashes?

Amy


-----Original Message-----
From: Doug Lochart <[EMAIL PROTECTED]>
Sent: Thursday, February 14, 2008 10:30am
To: General Linux-HA mailing list <[email protected]>
Subject: Re: [Linux-HA] resource script question (runlevel config)

>  First off you don't want to make sshd an HA service or you will not be
>  able to ssh to any node that is not the primary node. SSHD should be
>  running on all servers at all times.

Unfortunately the main app we are trying to make highly available is
an rsync based backup system.  Originally we were running rsync as a
daemon and tunneling over SSH.  We discovered a security hole in that
setup where a user could edit their scripts and change module names to
another client and gain access to their data because they were already
authenticated to the box via the ssh tunnel and rsync daemon has
access to all the modules.  Anyway we needed more control and better
security so we switched to running rsync within an SSH connection.
This starts an rsync process in daemon mode but it runs as that user
and can ONLY access modules in the user defined module in their home
directory.  We also added a lot of protection in an rsync wrapper
script that is run first to intercept and augment/protect/reject
commands coming in via SSH to limit a user from doing ANYTHING else
besides the rsync command.

To make a long story short SSH is the backbone of our app.  I did not
think about SSH not being available on the secondary node so that is a
concern especially for remote administration.

>  We start heartbeat at boot on all of our clusters and I can't see a reason
>  why you would not. You, of course, cannot start the services which you are
>  making HA as they are started by heartbeat when it runs.

Thanks.  Just trying to be clear.  I appreciate the tutorials as I
know they are a give back to others like me but when you are
struggling to come up to speed quickly (as is always the case) little
details ommitted make all the difference in understanding.

>  In any failover the nodes do know their roles when they come because they
>  talk to the cluster. How this is handled depends on your configuration. I
>  would advise using v2 configs and letting heartbeat manage the resources.
>  We have both v1 and v2 in our environment and while v2 is insane to set up
>  and get working just right it is worth it in the long run. The short
>  answer is that, in your scenario, server2 would indeed know that it is
>  primary.

Great.  So if V1 works why is V2 so much better?  I would like to get
this up and running in a test environment first under V1 so I can at
least get a grip on how it all works and do some testing.  After that
I may try to leap to V2.

regards,

Doug
>
> _______________________________________________
>  Linux-HA mailing list
>  [email protected]
>  http://lists.linux-ha.org/mailman/listinfo/linux-ha
>  See also: http://linux-ha.org/ReportingProblems
>



-- 
What profits a man if he gains the whole world yet loses his soul?
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems


_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to