[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-ha
Subject:    Heartbeat new directions (was: drbd question)
From:       "Luis Claudio R. Goncalves" <lclaudio () conectiva ! com ! br>
Date:       2000-03-30 20:52:52
[Download RAW message or body]

Hi!

On Thu, 30 Mar 2000, Alan Robertson wrote:

> But, it was pretty useful IMHO...

   I have no doubts. :)

> The "system" should:
> 
> 	Recover from single-node faults

   This almost done. It lacks (IMHO) the correct nice failback(tm)
behavior and the per link health stats that Marcelo is working
on. By correct nice failback I mean the sit and cry (tm) behavior or
whatever we all define as a good starting protocol.
   I'd suggest for the sake of simplicity that we create a file called
/ha-<node-name> (or so) when starting heartbeat and remove this file
when
performing a nice shutdown. This way the sysV script could identify a
node failback and perform all the operations we judge important.

> 	Detect and raise an error for multiple node faults
> 		(This is analagous to how 2-bit memory parity works.)

   We could also create a file /ha-<node-name> for each node in the
cluster and erase this file when we received a good bye (or
shutdown) message from a node.
   If we restart with two (or even more) /ha* files... there was some
problem in the cluster.
   It sounds good to me... any idea? 

> Most of the discussion seemed to assume these two.  The only extension
> over what has been talked about all along for drbd was that it should
> somehow "detect and raise an error" for the multiple node failure
> situation.

   It'd soved by the above idea. :)

						Hugs!

						Luis, the talkative one...

[ Luis Claudio R. Goncalves                  lclaudio@conectiva.com.br ]
[ BSc in Computer Science -- MSc coming soon -- Gospel User -- Linuxer ]
[ Fault Tolerance - Real-Time - Distributed Systems - IECLB - IS 40:31 ]
[ LateNite Programmer --  Jesus Is The Solid Rock On Which I Stand  -- ]


------------------------------------------------------------------------------
Linux HA Web Site:
  http://linux-ha.org/
Linux HA HOWTO:
  http://metalab.unc.edu/pub/Linux/ALPHA/linux-ha/High-Availability-HOWTO.html
------------------------------------------------------------------------------

[prev in list] [next in list] [prev in thread] [next in thread]