Roger Pena Escobio wrote:
How do people do it?
Here are some options you have:
1) Use a NAS/NFS box for shared storage - not really a
solution for high availability per se, as this becomes a
SPOF unless you mirror it somehow in realtime. Performance
over NFS will not be great even in a high state of tune due
to latency overheads.
2) Use a SAN with a clustered file system for shared
storage. Again, not really a solution for high availability
unless the SAN itself is mirrored, plus the performance will
not be great especially with a lot of concurrent users due
to locking latencies.
3) Use a SAN with exclusively mounted non-shared file
system (e.g. ext3). Performance should be reasonably good in
this case because there is no locking latency overheads or
lack of efficient caching. Note, however, that you will have
to ensure in your cluster configuration that this ext3
volume is a service that can only be active on one machine
at a time. If it ends up accidentally multi-mounted, your
data will be gone in a matter of seconds.
2b) Split your user data up in such a way that a particular
user will always hit a particular server (unless that server
fails), and all the data for users on that server goes to a
particular volume, or subtree of a cluster file system (e.g.
GFS). This will ensure that all locks for that subtree can
be cached on that server, to overcome the locking latency
overheads.
what about using a combination of 3 and 2b:
3b- split your users in a set of servers which use ext3 FS but are part of a cluster, the servers are really services of a cluster (IP and FS are resources of a CLuster Service) so, if a server fail its service can be migrated to another node of the cluster
let say for example
users starting with letter 'a' to users starting with letter 'h' will "assigned" to MailA , users from 'i' to 'z' will be assigned to MailB.
MAilA --> ipA and filesystemA
MAilB --> ipB and filesystemB
Cluster ServiceA will have resource ipA and filesystemA
Cluster ServiceB will have resource ipB and filesystemB
and ServiceA will be configured to run in nodeA, while ServiceB will be set to run in nodeB of the cluster, but will be set to failover to nodeC (standby server)
the hard part of this is how to balance the users between MailA and MailB (and MailC , D, E). Changing the value of "mail_host" in user attr (if using a Directory Service) and moving user's email from one filesystem to another
this is just food for the brain, the scenario could be as complex as you like :-), but definitely is no good idea to have GFS for mail servers if the clients can connected from multiple sources and dont have a "proxy" to tunnel all request for same user to same backend server.
We use a mail system called CommuniGate Pro here in cluster mode. The
storage is mounted onto all four of our servers via NFS. Any account can
be opened on any of those four servers, but the cluster prevents that
account from being opened on more than one server at a time. It forwards
connections internally to the correct server. There are probably other
mail servers out there that have internal clustering functionality that
deals with these problems.
--
Mike Cardwell - IT Consultant and LAMP developer
Cardwell IT Ltd. (UK Reg'd Company #06920226) http://cardwellit.com/
Technical Blog: https://secure.grepular.com/blog/
--
Linux-cluster mailing list
Linux-cluster@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/linux-cluster
