Disk Backup

Day before yesterday we had a disk crash on one of our internal mailservers.
The metadb failed on a raid 1 mirroring disk, problems should not have been so big.
However the synchronisation worked not so well. Solaris keeps a record of each disk and in which scsi slot it is located. Therefore both disks synced with eachother making the filled disk sync with the empty one, moving all data to Lost+Found. No way everything would fit in there be easily restored.

2 weeks before I had to restore a mailbox of one of our users, resulting in a 2 day tape restore.
The old LTO drives are slow. I did not want to spend another 2 days restoring mail for users in the future. So I wrote a little script that makes by default a UFS level 1 dump to harddisk, unless it does not see a level 0 dump. Making level 1 backups is quick and is done twice a day outside office hours.

It was this disk backup that saved us on the second day, ufs restoring all users email in just 2 hours (level 0 and 1). There only remained a small issue with the mail service. This isn't your typical sendmail but a special mail service used by some of the larger ISPs. Some users where left with broken mails FETCH issues with imap and a broken POP2 fetch rfc822 implementation remnant from netscape, which microsoft Outlook implemented as well.