[DRBD-user] rolling upgrades

[Weyer, Wolfram]

Hi,
have there been thoughts about upgrading drbd in a failover-cluster without
shutting down the service?
 
I am thinking of this scenario:
 - put standby server to offline
 - upgrade drbd on this node
 - bring standby server online again
 - resync the newversion-standby with oldversion-active
 - failover
 - then upgrade the former active
 
Currently there is a piece of code that prevents different-version-drbds to
communicate in drbd_receiver.c, which means that a bugfix-upgrade for drbd
causes a significant outage for the server. This hurts in an HA application.
Solutions like md in a shared storage environment can deal much easier with
this.
 
First I can think of bugfixes that have no effect on data-structures and
data-flow over the network, there is no need to force the versions to
incompatible then. Then checking with a new define similar to API_VERSION
for drbduser, which does not increase for these changes would make it
happen.
I doubt there is a change to implement rolling upgrades from 0.6.x to 0.7
but having this in mind for the 0.6 maintenance releases would help a lot.
 
/Wolfram
 
 

Wolfram Weyer 
Staff Engineer 
Force Computers GmbH 
- A Solectron Company - 
Lilienthalstr. 15 
85579 Neubiberg/München 
Germany 
Tel.  +49 (0)89 60814-523 
Fax +49 (0)89 60814-112 
Wolfram.Weyer at fci.com 
http://www.forcecomputers.com <http://www.forcecomputers.com/>  

--- This e-mail is confidential and privileged. If you are not the intended
recipient please accept our apologies; please do not disclose, copy or
distribute information in this e-mail or take any action in reliance on its
contents: to do so is strictly prohibited and may be unlawful. Please inform
us that this message has gone astray before deleting it. Thank you for your
co-operation. ---


 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.linbit.com/pipermail/drbd-user/attachments/20040216/704b3588/attachment.htm