On 01/-10/-28163 02:59 PM, Graeme Fowler wrote:
+1 for rsync. In fact +++ for rsync.
They joyous thing about rsync is that you can run it on a live system,
limit the bandwidth it uses, so you take a clone of the "live" files
without turning the system off.
Do that several times over a few days, then when you finally want to
switch you stop the main CGP service and run a "catchup" copy which will
take far less time than the initial one (and you can run it without any
limits applied). When it's finished, your filesystems are in sync, so
you run up the new server(s) and away you go.
Regarding OS support - we've been lucky and haven't needed it. However
we are having nightmares with the transition from Sun to Oracle with our
many support contracts for other systems; as a new customer you would
probably get better service.
Graeme
To add an opinion on several posts.
Definitely use RedHat or CentOS over Fedora on the server itself. Fedora theoretically requires two OS upgrades a year. Since you mentioned that you were moving to a cluster, I'd definitely consider one or two with a RedHat license, and any additional on CentOS. If you mix, just make sure that you are deploying the same Major version. (CentOS has not released v6.0 yet. RedHat has 6.1 in beta. You'd be better off sticking with v5 for both in the short term.) The RedHat license and support will probably only be necessary if you have hardware related issues. The need for this is reduced a bit since you are deploying EMC storage. (You can rely on them for storage related support, which seems the more likely situation for a mail cluster.) Just don't go cheap on your HBAs. If you're buying name brand servers from a vendor that provides Linux drivers/support, you wouldn't be risking as much from a support perspective. The one time I really needed support from a Linux vendor was with a NUMA scheduling issue in the kernel which caused the system to hang. If you're using hardware that has been around for a while, you may not need the RedHat support. (Unless you're interested in taking RH training, then buy the license.)
I can't speak about Oracle's Solaris support but definitely agree about ZFS filesystem in a local storage situation. Don;t have experience with the comparison when clustering with EMC SAN behind it.
I also agree that several passes with rsync or restore from backup followed by rsync while CGP is shut down is the way to go for minimal downtime. Google "rsync performance tips" to get the most benefit. EMC (or their ISV) may also have some other suggestions for direct storage to storage migration.
#############################################################
This message is sent to you because you are subscribed to
the mailing list <
CGatePro@mail.stalker.com>.
To unsubscribe, E-mail to: <
CGatePro-off@mail.stalker.com>
To switch to the DIGEST mode, E-mail to <
CGatePro-digest@mail.stalker.com>
To switch to the INDEX mode, E-mail to <
CGatePro-index@mail.stalker.com>
Send administrative queries to <
CGatePro-request@mail.stalker.com>