The read and write performance of a storage device is low. Provided that no obvious service load increase, the service performance degrades.
Product and version information:
1. Replace the disk. Do not compromise service data during slow disk replacement.
2. Recheck by the same method to confirm whether there are some other slow disks.
1. Log in to the storage device through the serial port or the SSH.
2. Run the debug command to go to the debug mode.
3. Run the iostat –x 2 command to view the read and write performance parameters of member disks: svctm (average service time) and util (utilization).
Device: ...... await svctm %util
sddh ...... 30.66 8.79 61.50
sddg ...... 1437.00 18.54 100.10
sddf ...... 37.17 9.41 60.20
sdde ...... 21.03 9.05 56.10
In this case, the statistics unrelated to the fault are omitted.
4. In the statistics, the disk that has a relatively larger svctm value than other disks and whose util is close to 100% is a slow disk, such as the disk sddg in the previous example.
Therefore, we have the following conclusion:
The overall read and write performance of a storage system degrades if there is a slow disk.
1. The command iostat can be used for collecting the average service time and utilization of member disks only when the storage device performs read and write operations on mapped LUNs.
2. If you run the iostat command for the first time, add –d behind the command to obtain precise statistics.