Assume a 2-drive RAID-1 array on an HP DL 380 G7 server with the P410i raid controller. If the controller determines that one of the two drives has failed, I assume that it will automatically start using only the single, still-functioniong drive and that operations on the server won't be impacted by the failure. My question is how the controller card signals to the operating system (Windows server 2008 SP2) that a failure has occurred. Is is possible to have user application code be informed that a failure has occurred so that it can pass that information along to a remote user? If a failure has occurred, I'd like to have our UI put up an error dialog or icon to inform the user that this has happened and for them to initiate a replacement of the failed unit.
5 Answers
I know that machine very well (and the G5/G6 etc. versions) - basically R1 uses BOTH disks when they're happy, it's not like an active/passive setup - both disks are racing for reads against each other and the array isn't fully written to until both disks have the same data. What you need to do is install the latest Proliant Support Pack (let me know what OS you're using and I'll point you to the exact HP.com page), this included disk drivers that will alert the OS event queue when there's a prefailure/failure so that you can deal with the problem right away.
Hope this helps.
- 101,299
- 9
- 108
- 239
This is going to be hardware dependent. Generally, the RAID controller will completely "hide" a disk failure from the Operating System, at least as far as normal usage goes, so the OS doesn't need to know, worry, or care about a disk failure. It just keeps on chugging along as if everything is fine.
Typically, you'll need to use some sort of special tool or utility, generally supplied by your Server or Storage vendor, in order to interrogate and detect hardware problems underneath a RAID setup.
Look for RAID controller management tools and software at (in this case) HP's website, probably bundled with or near the drivers for the box. They should offer something that you can monitor or that can alert you to hardware issues.
- 9,128
- 2
- 32
- 44
Doesn't HP supply utilities to monitor the array? I would check the HP website for P410i raid controller utils.
- 488
- 2
- 8
If you install all the software that came with the machine it should include a monitoring tool which presumably supports informing local users or sending emails when issues occur. If you don't have the relevant install sets they should be findable on HP's support site which is usually pretty good at having all the relevant drivers, utilities and documentation.
There is often a hardware indicator too in the form of an LED on the front of the machine. Again you'll find details on this in the documentation (which will be online if you don't already have a copy to hand) and also marked on the map that is usually printed or glued on the inside of the top of the server's case.
- 22,754
- 45
- 67
If you wanted your application to throw up a warning to users that there is a failed drive, you will have to have your application poll for the array status. How you do this will of course depend on your application and how the array software can report failure.
However, I'd recommend that you set up a monitoring system which actively polls for this condition and alerts someone when it happens. One issue is that this sort of polling, if it fails will tend to fail by alerting you ("can't get array statistics"), rather than potentially failing silently in your application (the array program isn't reporting "failed drive", so I won't throw up the alert, even though it's also not reporting success). Plus this means that someone knows it's their job to take action on it, rather than all application users seeing it and thinking someone else will be looking at it.
Monitoring systems are very, very good things for handling this.
But, you're ahead of the curve that you're making sure you set up array monitoring. Many people set up RAID and then forget it until the SECOND drive fails. :-(
- 10,720
- 3
- 25
- 28