disk I/O on DL360 G-3

[Chris Rice_1]

HELP,
I have a DL360 G-3 running a Raid 1 array with 3 logical drives. (C, D, and E) I am experiencing an issue where randomly throughout the day the disk I/O on the D drive will max out at 100%. I am not having any issues on the C or the E drive. The server is configured with two 36.4 HDs.

Any help or suggestions would be very much appreciated.

[Prashant (I am Back)]

Hi,

I am not clear about the problem but just as suspecting we can run hdd identifier utility from hp site to verify that hdd we have is not from faulty lot as stated by HP.

Regards,
Prashant S.

[Chris Rice_1]

Well, the exact issue is that randomly throughout the day, the disk I/O will spike at 100% and will stay there for approximately 1-3 minutes then it will go back down to a normal level. We have been looking at traces and it appears as if something is doing a "dir" in a command prompt or something like that because we see a large amount of scans through the directory structure on the server. I know this may sound like a software or user problem, but any input would be a big help. We have removed almost all user access to this server to help troubleshoot.

[e4services]

May be bad sectors on one of the drives is causing the problem. Takes some time for the bad sectors to be marked and data recovered.
Just a thought

[Uwe Zessin]

Could it be a built-in indexing service?

[SAKET_5 · Accepted solution]

Hi Chris,

A couple of suggestions:
1. Are you running the latest firmware on the drives? I know if it is a firmware issue then you should (theoretically) experience this issue on all the logical drives on the same disk. However, no harm in upgrading the disk firmware.

2. What do you use to monitor disk I/O on the drives, just perfmon (?), my point is are we sure no false positives are being obtained?

3. You could run PAT (Performance Assessment Utility) from HP to confirm the logical drives read/write performance on drives.

4. Could you run chkdsk from command prompt to see if if you are actually developing disk errors?

5. Could you try chkntfs on the affected volume? It should tell you straightway is the volume is dirty or not?

6. Are you running any Antivirus scans around the suspected drive or any intensive disk I/O occuring on the suspected drive?

7. Have you placed paging file on this drive?

8. Is your volume heavily fragmented?

Hope this helps and look forward to your responses,

P.S Don't forget to assign points:)

Regards,

[Chris Rice_1]

We have found what the issue is. We are using BMC patrol to monitor this server. We have found an agent that is doing a check that is causing it to scan every subdir. in the structure of the D: drive. Since this server is being used for files, there is a very large amount of dirs. so in turn, it is maxing out the I/O. We have disabled the agent and our group that does monitoring is researching a different way to handle this.