Stability Problem and New RAID-1 System

Apr 27, 2007 | 1 minute read
Share this:

Tags: LVM, RAID, Crash

Just after the big update and new infrastructure installation during the past month, the server encountered stability problem on a regular basis (sometimes crashing more than one time a day). At first though, we noted it may be caused by the new processor architecture, a less mature FreeBSD distribution than the well known i386.

But the problem appears to be specifically related to I/O, especially when the input/output are very intensive, i.e. file system dump(8), file system snapshot mksnap_ffs(8) or big tar(1) or cpio(1L) archive transfers.

After days, we didn't succeed to clearly isolate the source of the problem, but decided to quickly put the system under RAID-1 system management (disks mirroring) and configure the excellent gmirror(8) to do the job. So, i switched from an old hardware RAID mechanism, to a pure LVM one. Hope we will be able to find something new without disturbing the ThNET services anymore.

More on the involved manipulations to put the system under the gmirror(8)control on a later post. Stay tuned.