Quote:
Originally Posted by bhwong I know we can set a secondary MX record that will take over the primary MX when the primary email server failed. The question is, when the primary email server is up, there will be email getting into the secondardy email server during the period the primary email server was down, so how do the email client download email from the secondary email server without having to amend their POP?
Not to mention that amending POP will lead to redownloading of all email again. Luckily Zimbra has the option to select download email from now on. A very useful feature.
How do you guys setup a backup email server when the primary failed? Do you use MX record with lower priority too? |
We configure a separate, off-site plain Postfix server to serve as a backup MX for our production Zimbra hosting farm.
If spammers try sending to the backup MX while Zimbra is up, we don't care, the backup MX cleans the mail just like Zimbra and sends it on if clean immediately to the Zimbra system.
When the Zimbra system is down for maintenance say, other servers trying to send to Zimbra see that Zimbra is unavailable and send email to the backup MX.
The backup MX is configured to try to forward all legitimate email to the Zimbra server farm; if Zimbra is unavailable, the backup MX retries periodically.
To eliminate backscatter and ensure the backup MX only accepts email for valid recipients on the Zimbra system, we run a cron job on the Zimbra system that extracts all valid emails from the system. The output is sent to the backup MX, which runs a separate script that looks for any changes (domains/email addresses added/deleted) and updates the backup MX server's Postfix configuration accordingly and then reloads Postfix.
DRDB it seems from the posts here can also work; but the method above and DRDB are really designed to handle two different Disaster Recovery use cases.
I'd start with the use cases your management needs to address, and then the appropriate solution(s) should be straightforward to determine.
Hope that helps,
Mark
__________________
___________________________________
L. Mark Stone, CIO
"Uptime. All the time."
477 Congress Street | Portland, ME 04101-3431 | (207) 772-5678
proactive maintenance and monitoring | technology consulting
Zimbra groupware | EMR implementations | private cloud hosting