Step-by-step upgrade from DRBD 0.7 to DRBD 8
This step-by-step documentation worked fine for my migration. The only problem I got was the fact that some of my parameters used in my drbd configuration were deprecated. Now I use the example configuration provided by drbd package.
After the migration I wanted to perform minimal tests on the newly migrated system. First of all I tested main features:
- Is it possible to set on both systems the drbd status as primary and to mount the file system?
- Do both systems reconnect drbd device after a full reboot successfully?
- Is heartbeat possible to control drbd device and file system, on both cluster nodes?
- What happens when the connection to other node got lost?
With following command it is possible to write random data to destination file:
~# dd if=/dev/urandom bs=1M count=51200 of=/data/50GB-random.bin
if=/dev/urandom defines as source file /dev/urandom
bs=1M defines that dd reads one megabyte of source
count=51200 orders dd to read source 51200 times
of=/data/50GB-random.bin defines the destination file
While the upper command was working I tested the speed of this write process with iostat and iftop. iostat easily shows the I/O bandwidth of the local hard drives. iftop shows the network bandwidth of defined network-adapter.
I know that my tests are not perfect because there are no tests of concurrent file access or similar like this. It is also strongly recommend to do complete new failure and recover tests, but these tests I will do later.
Posts
