Hey

I run RAID 1 on our servers. As a test ( on a test machine), I pulled out the power from on of the mirrored disks while the system was running.

As you would expect (and as I expected and was releaved to see) the system didn't care. It just kept running.

However, when I looked at

# cat /proc/mdstat

it still showed that all the RAID devices were present. It wasn't until I rebooted a few minutes later, that the machine updated mdstat to show one of the HDDs was missing.

If I hadn't rebooted, does mdadm periodically check that both HDDs are still there (so taht it could hot add the replacement)?

Thanks
Hamish

The man page for mdadm indicates that the default poll interval is 60 seconds, assuming that you have started the md monitor. The RAID array must be declared failed, which it probably won't be until a write attempt (since reads can be satisfied by either drive in a RAID1 array).