MAIL9 Disk Maintenance

[Danl]

I wasn't aware that anybody was having issues on mail9, do you have a support ticket in?

[mrwill]

Originally Posted by JoshK

I'll attempt to answer all the questions in one shot.
You cannot run a backup copy of mail. It is volatile storage. Backing it up would be futile as data on the hard drives literally flows through storage it is not kept for long periods in most cases. The speed loss of writing backup copies would degrade mail service significantly. As sits we put the highest RPM drives readily available in these machines just to keep access times down. We do have a backup of the OS drive, but that drive sees little wear and hence rarely fails.
We assess hardware when we notice there is a problem. Would you rather I wait till the start of your business day, but then have the server down for two or three days while we configure a new one? This was not scheduled, I saw errors, took it off line in hopes of a quick fix and instead saw hardware issues. While the machine was down, we slid a new machine complete in it's place. This was to prevent future problems.
As to location, we do schedule most things in the middle of the night in our time zone. That is because statistically there is less traffic. We watch the total numbers and attempt to schedule all tasks that might incur machine downtime in the times of least traffic. When possible we schedule planned take downs for the middle of the night on Saturday here, because the total worldwide traffic is the lowest then and it is not buisness hours anywhere.
To repeat though, this was not a planned take down. The machine was displaying errors, and when we checked the disks the error quantity was too high. When I ran performance tests the disk was failing. At that point it was replace it now and have limited downtime, or risk loosing all mail and possibly days of downtime. I must do what is best for all customers involved. A small delay now is better than days of downtime and lost data later.
Currently new hardware is in place, we're copying all the data off the old drive to the new drives. We expect it to be fully functional soon. It may be functional for some customers now as the directories are restored.

Thanks for the update..but that still doesnt explain why you don't run critical mail servers in a cluster environment with a separate information store? That would of been NO down time? I assumed that was already the case?

[Brangwyn]

A configuration like that isn't supported by HSphere AFAIK.