|
Hi,
this is exactly the same Problem I have bumped into: How do I replicate the
data on my Fileserver in realtime to another "hot spare" machine ...
The "solutions" I found so far:
a) use Coda as filesystem
AFAIK Coda allows realtime replication. Coda is still not ready for
production use though so I am no considering using it
b) have both Fileservers access the same Disk-Array
Set up a "dual-homed SCSI" system as described in the Linux-HA Howto so that
_both_ fileservers access the same disks (which are mirrored via RAID 1). If
one fileserver fails the other mounts the disk-array. This will only make
sense when using a journaling filesystem though. There are lots of
"in-development" journaling filesystems for linux but AFAIK none of them is
"stable" yet (ReiserFS, XFS, JFS, ...) If you are not using a journaling
filesystem the fs-check on a takeover of the disk-array will take several
minutes which is not acceptable in my case
c) use "slow", not-realtime replication
this can be done easily once per hour or so via rsync.
Are there any more options ? I have now spent several hours doing "research"
on the web but haven´t found a good solution for this problem though ...
Greetings,
Jochen
From: root <ian@xxxxxxxxx>
To: <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Sent: Monday, April 24, 2000 3:22 PM
Subject: Simple Replication...
> I have looked at your software documentation. I am interested in
> running your software, however, is a "Load Balancer" node needed? I
> would like to run two servers of the same hardware. Should one fail,
> the other will take over. Will your software be able to run like that?
>
> Most of what I have seen in the lines of server replication is simply a
> timed script to replicate the data and an IP Auto Take over system. I
> would like to run live replication.
>
> Any clarification in the matter is GREATLY appreciated.
>
> Thanks,
> Ian Langdon
> Interface Technologies
> Rescue.to
>
>
>
>
>
|
