HA mailserver with Poste.io (possible?)

[Former user]

Hi all,

I am currently running a poste.io pro instance for a company with around 80 mailboxes. Everything is working fine so far. Unfortunately last week we had two increase the storage on the server which has led to an one-hour downtime.

As you can imagine, the employees were not very enthusiastic about it. To avoid this kind of frustration the company bought another exact same server with the same capacity. I’ve already set up a failover IP behind the two servers. In theory when a server needs to get maintained all the traffic is getting routed to the backup server.

At this point the only thing where I’m struggling on is the synchronisation of all mails, accounts and settings between the two server. It does not need to be in real-time, maybe once an hour to avoid overhead.

I tried it with rsync and a cron to completely sync the poste.io data directory to the backup server while poste.io is running. Everything worked great (unless your are not changing some critical poste.io settings / custom Haraka plugins). But in case of a downtime by the mainserver I don’t really know how to transfer the new mails, came to the backup server during this period, back to the mainserver when it’s on again? I don’t want to replace all files with the ones from the backup server when everything is up and running again, since the main server might have some new mails before it went offline.

Is there any recommendations how to create a high availability Poste.io failover server with two separate hard disks?

Thank you

[Scott MacDonald]

Are you using physical or virtual servers?

[Scott MacDonald]

If using virtual servers you could mount a second drive to house the poste.io mailserver /data folder. Then disconnect the drive before performing maintenance and connect it to the failover VM.

Another option is to perform server management/maintenance afterhours when there is little impact to your end users.

Additionally, if virtual using a second drive for storing the /data folder means you can easily create another larger drive and mount it and copy over the data to it. Might find it faster and easier than trying to extend a drive from within Linux.

[SB]

Would a simpler solution not just be to use a backup MX which forwards to the main one in the event of it being unavailable? Even using poste.io and setting the domain as a bin to forward all possibly?

This means you wouldn't lose any email coming in while the main server is offline, and when ti comes back up all new emails received would be delivered, plus there is no data to sync.