What strategies do you use to monitor the health of your server hard disks? Do you schedule an automatic chkdsk with reporting? How are you alerted of failures?
-
1What OS? I could ramble on about smartctl but if you're using Windows it may not help much.– MadHatterFeb 26, 2011 at 17:50
-
Was the question just severely edited? Could have sworn you wrote windows...– CampoFeb 26, 2011 at 18:06
-
Windows - various versions 7, server 2008, xp– SurajFeb 26, 2011 at 18:06
-
3chkdsk reports corruption, not disk health. Once you've found corruption, you already have likely data loss.– jgoldschrafeFeb 26, 2011 at 18:06
-
thanks all for comments. I'll look into RAID via the mentioned tools. For non-RAID, can someone propose a specific strategy? (i.e. install tool ABC for SMART monitoring, set it to run every day with xyz parameters. schedule a CHKDSK task every day with xyz parameters, etc.). The more detailed the better. ANYTHING you propose will be better than what I have now which is nothing.– SurajFeb 26, 2011 at 18:14
3 Answers
There is actually a lot to monitor within the Disk Subsystem:
- Filesystem Health: either chkdisk or fsdisk (The File system is not the same as the health of the physical array -- but of course physical problems will generally lead to FS corruption).
- RAID Health: Includes state of whatever raid configuration and the battery on the raid controller. For example has a disk dropped out of the array, is it rebuilding?
- Predictive disk failure for each disk: See S.M.A.R.T
- Operating Temperature (not particular to disks but important).
- Up to date on firmware / drivers
You also probably want to monitor the performance of the disk subsystem:
- Average time taken of read / write operations
- The amount of reads / writes
- Disk Operations Queue
A system like Nagios has various plugins and can alert based on these failures. Often you need a utility like Dell openmanage or MegaCLI to monitor the status of RAID via SNMP. The tools can also alert you themselves usually but that is not centralized.
Most modern hard disks have a built in monitoring called S.M.A.R.T. (wiki)
There are many tools for your OS that would be able to monitor that.
For windows look into
Active@ Hard Disk Monitor Freeware just as a starting point
Your RAID controller may have built in S.M.A.R.T. monitoring as well so look there if this is for a server.
-
what is the overlap between SMART capabilities and chkdsk capabilities?– SurajFeb 26, 2011 at 18:06
-
Event Logs, WMI, SNMP.
Most server-grade RAID controllers will include software to enable one or all of those in Windows and other operating systems.
Email alerts are often available as well.
We do weekly or daily consistency checks on our RAIDs, and do regular backups.