Re: [Exim] SMTP Cluster - Shared Spool?

Góra strony
Delete this message
Reply to this message
Autor: Suresh Ramasubramanian
Data:  
Dla: Jawaid.Bazyar, exim-users
CC: bazyar
Temat: Re: [Exim] SMTP Cluster - Shared Spool?
(Jawaid.Bazyar@???) [Tuesday, August 13, 2002 1:48 AM]:

> Has anyone done a performance analysis of queue-handling on different
> filesystems; e.g. Ext2 vs. Sun UFS vs. ReiserFS vs. XFS?


Reiser is not bad at all, for a queue. Theo spoke about using a hacked
version of exim (with fsync and other calls removed) and XFS in a post
sometime back on exim-users.

> Do any MTA's out there take a custom-filesystem approach, like some
> of the UseNet/NNTP servers do?


Some MTAs (InterMail for example) use a SQL db for the message store, but
the spool is generally a conventional filesystem.

> The reason I wanted a central file store is that high-redundancy file
> stores are relatively expensive to build, operate, maintain, etc.
> Also, if the only thing that server is doing is serving files with
> NFS, they are also very reliable.


For your spool? Locking is going to be hell over NFS.

For your message store? Fine, enjoy life with maildirs over NFS, or a SAN.

> In either of these setups you'd still have a nice fat traditional RAID
> file server for user mailboxes (maildirs), and the spool front-ends
> could then deliver directly to that filesystem over NFS.


If you do want to implement SAN, put your mailstore on a SAN for more
reliablity. RAID hard disks in a single box have a nasty tendency to fall
over and die sometimes.

I'd just suggest that you leave your mail spools on separate machines, with
fast SCSI hard disks, a fast filesystem and ensure that mail moves out of
there into other machines for onward transmission or final delivery ASAP
(use mailertables if necessary).

    -suresh