[Linux-cluster] Re: generic fencing, stonith, etc [was: gfs and fencing]

Birger Wathne Birger.Wathne at ift.uib.no
Tue Feb 21 12:51:36 UTC 2006

Lon Hohberger wrote:

>Certainly, the 'ssh' agent could automatically recover in a few more
>cases than the 'manual' agent ('meatware' in Linux-HA lingo).
And with multi-level fencing, ssh could be tried first, then the 
meatware could be stirred...

>Someone mentioned recently (elsewhere) that it would be great if we had
>a fencing agent which just called some user-specified command to do
>things (and could substitute variables if necessary, like %p ->
>password, %l -> login, etc.):
I agree that this would be very nice :-)

>(a) I do not think this is not a supportable solution, due to the
>limitless array of possible configurations for hardware we have never
>heard of.  All support for this agent would likely be limited to this
>mailing list and bugs in the actual agent itself.
I don't think that would be a problem. Most (maybe not all) admins 
should be smart enough to understand that you cannot support the command 
used within this agent even if the agent is supported.

>(b) Rather than developing a proper fence agent for their particular
>hardware, people will use this as a means to an end, which will not
>improve the linux-cluster project as a whole.  That sucks :(
Is there any good documentation on how to make a fencing agent? What 
kind of info the agent can get from the cluster service, etc... Would be 
great! I would like to automatically find a list of scsi-devices shared 
between 2 systems in a 2-node cluster so I could hack up a fencing 
device that used scsi reservation. Combine this with the stonith option 
in the scsi driver and I would have a nice fence for cheap clusters. If 
I have to manually configure the list of devices it would be easier to 
use the proposed fence agent to run some existing command-line tool.

>So, given that it is not up to me, who else wants this (I'm talking to
>you lurkers out there).  I am waiting for the flaming mantis to burn me
>for even suggesting such a thing...
I certainly want it. It would fill an immediate need to get up and 
running with better fencing than the manual one. Something as simple as 
ssh would have a better chance at getting the system quickly back to 
service than waiting for the meat...


More information about the Linux-cluster mailing list