Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
Milind Dumbare <milind at linsyssoft.com> writes:
> On Wed, 2006-08-23 at 23:32 +1000, Daniel Pittman wrote:
>> Milind Dumbare <milind at linsyssoft.com> writes:
>> > On Wed, 2006-08-23 at 18:59 +1000, Daniel Pittman wrote:
>> >> Milind Dumbare <milind at linsyssoft.com> writes:
>> >>
>> >> > Sorry for very may be stupid question. But do we need heartbeat
>> >> > running on both nodes in 0.7 or 0.8pre3.
>> >>
>> >> No. You don't need heartbeat running at all for DBRD to work.
>> >> What you do need heartbeat for is to make DRBD *useful*.
[...]
>> If you want the cluster to actually *do* something when a machine fails,
>> or whatever, then you need something more than DRBD.
[...]
> Ok, My question is I have heartbeat stopped at my system.
Sure. Presumably you don't have anything else doing the same sort of
job either, like the 'keepalived' or 'vrrpd' packages.
> Now with DRBD(0.7.X), if my primary machine fails then my secondary
> should be primary, right?
Wrong. The secondary machine will stay secondary, although DRBD will
notice that the peer has gone away and reflect that.
> Will this be affected by heartbeat not being ran.
...yes and no. No, in that DRBD will still perform correctly.
Yes, in that DRBD will not become primary, and the file system on it
will not be mounted, and so forth. Without something like heartbeat
DRBD is, effectively, a passive component.
> Basically what I have observed with Heartbeat on my system is when I
> start heartbeat daemon on both nodes (panini, xenon), one of the
> machines(panini) creates one more profile to my eth0 (eth0:0) with ip
> address of other machine(xenon). So all my ssh connections to xenon
> get disconnected, and this also causes WFconnection in DRBD.
>
> Infact this question should be on linux-ha list but as it is with DRBD I
> have posted it here also.
OK. The short answer is: you don't need heartbeat, but you need
something that does the same job.
You should take your question to the heartbeat list and ask there,
because you need to configure heartbeat to work as you expect.
Er, if it isn't too bold, though, I would also strongly suggest you have
someone with solid experience in high availability design look over your
system[1] -- because if you need to ask that question then you probably
have a bunch of places where you system will *not* fail over correctly
when a machine fails.
Regards,
Daniel
Footnotes:
[1] As in, hire a contractor.
--
Digital Infrastructure Solutions -- making IT simple, stable and secure
Phone: 0401 155 707 email: contact at digital-infrastructure.com.au
http://digital-infrastructure.com.au/