RAID1 Test Suite for LTP

written by Fei Fei

fei.fei@intel.com

Testcase Number  Test Procedure  Expected Results 
1 Verify that disconnecting the power from disk1 will keep the system running healthy
  • Boot system with two disks connected 
  • cat /proc/mdstat shows the two disks working in raid-1 mirror 
  • Remove power cable from disk1 while the system is running. 
  • Ensure the system is still responding, and you can still read and write to the disk. 
  • Also, ensure an error message in /var/log/messages indicating one of the disks is off-line. 
  • If you have event logging installed, then look for event logger error (event_type=1021) using evlview -filter 'event_type=1021' 
2 Verify the system can boot from disk2 (the only connected disk)
  • Before you do this step, make sure that lilo is set to boot from /dev/mdx device and not the sd device (boot=/dev/sdx).
  • Boot system after test case 1 
  • Verify the system reboot all the way with no problem. 
  • System responds to I/O commands after boot 
3 Verify the system boots from disk 1 only (disk 2 disconnected)
  • Before you do this step, make sure that lilo is set to boot from /dev/mdx device and not the sd device (boot=/dev/sdx). 
  • Take the system down (normal shutdown) 
  • Disconnect the power from disk2 and connect it to disk1. 
  • Verify the system reboot all the way with no problem. 
  • System responds to I/O commands after boot 
4 Verify that re-connecting the power to disk 1 and disk 2 will keep the system running healthy.
  • Shut down 
  • Re-connect the power cable to the disk was removed (connect both disks). 
  • Bring the system up 
  • Verify the system comes up healthy 
  • The OS sees both disks using "fdisk -l" 
  • One of the raid disks is attached (cat /proc/mdstat) 
5 Verify that you can bring the system to a healthy raid state after re-connecting both disks
  • After case 5, run raidhotadd on all partitions of the inactive disk. (e.g. raidhotadd /dev/md0 /dev/sdb1) 
  • cat /proc/mdstat shows remirror is in progress 
  • Once completed, cat /proc/mdstat will show two disks are active. 
  • Verify the system still healthy (can write/read to disk) 
6 Verify that inserting a new disk will keep the system running healthy
  • Take system down (normal shutdown) 
  • Connect a new secondary disk (disk 3) to the system along with one of disk 1 or disk 2. 
  • Boot system 
  • The OS sees two disks using "fdisk -l" 
Note: If hot-insertion feature supported, skip test case 6,7,8,9 
7 Verify that you can bring the new disk to a healthy raid 1 system.
  • Partition the disk to the same size of the active disk 
  • Run raidhotadd on each partition 
  • Check /proc/mdstat shows the new disk is being added to raid-1 
  • All partitions will copy with no problem 
8 Verify that the system can reboot after copying all data to disk 3
  • Update lilo (/sbin/lilo) 
  • Reboot the system with both disks 
  • Verify the system reboot and comes up with two raid disks are attached. (cat /proc/mdstat) 
9 Verify that the system can reboot from the new inserted disk only (disk 3)
  • Take the system down (normal shutdown) 
  • Disconnect power from disk 2 or 1, and leave the power connected to disk 3, then bring the system up 
  • Verify the system reboot all the way with no problem. 
  • System responds to I/O commands after boot