Tim Alberts wrote:
John Summerfield wrote:
Are you serious? You are running mission-critical applications on the
second least reliable software offering from the RHL family?
I'm guessing you don't work in advertising for Red Hat, Fedora, or Linux
in general?
I don't.
Use good hardware, good software (RHEL or a clone), IMAP and not POP3,
and use one of the reliable RAID (1, 4 or 5) choices for your mail
(and other critical data) storage.
Been using POP3 forever with no problems (except some Mac problems with
You do have a problem, your mail is scattered all over the place. imap
keeps it on the server.
Dovecot). Have RAID1 software setup for years as well. Keep getting
drive failures and looking back into hardware RAID with high quality
equipment (as mentioned before).
Even if a dodgy Fedora software update doesn't get you, you still have
to contend with frequent upgrades of the software.
Yes, I've dealt with 'dodgy' updates and config files being lost by auto
updates, and bug fixes that mess things up that worked fine. For the
price, it has been acceptable (at least to the people in charge of the
purse strings).
Price a problem? CentOS is free of charge. It also costs less.
Note that RAID _can_ include a network block device (nbd or enhanced
nbd drivers), and drbd also provided RAID1 over a network, and is
tolerant of breaks in connectivity.
note that LVM can provide hot backups.
One trick I've hard of is to define a firewire drive (presumably USB
or other hotplug drive) would do as part of a mirror pair. Backup goes
something like this:
Plug it in
Resync.
Detach (I don't recall the fine details here)
Unplug.
Used removable drive trays for rsync backups without the RAID. Now we
got backup systems that are rsync backed up and ready to run in
failures. However, any email that was delivered between the last rsync
and the failure gets lost temporarily or permanently. Hence my idea for
NFS mount to quality RAID1
NFS has its own problems. enbd and drdb are better than rsync.
Google for terms such as "reliable linux" "high availability linux"
"linux cluster" etc for more details.
Google'd and Yahoo'd...seen hundreds of ideas mostly based on
heartbeat. In fact I most recently was looking into Red Hat's Global
File System and clustering:
http://www.redhat.com/gfs/
http://www.redhat.com/cluster_suite/
Trying to figure out how Red Hat is accomplishing these things with open
source, or if they are adding their own proprietary background stuff.
Look at CentOS. If it's in that (I believe it is) then it's OSS.
All-in-all, fun discussion, but completely off topic from my original
post and still doesn't answer my question.
I'm more concerned with the underlying problem than with your proposed
solution.
Do you want the best solution?
--
Cheers
John
-- spambait
1aaaaaaa@xxxxxxxxxxxxxxxx Z1aaaaaaa@xxxxxxxxxxxxxxxx
-- Advice
http://webfoot.com/advice/email.top.php
http://www.catb.org/~esr/faqs/smart-questions.html
http://support.microsoft.com/kb/555375
You cannot reply off-list:-)
