Running FC3 to 9, using mdadm, I just don't trust raid1.

Is there a way to verify that the two drives are perfectly in sync
in real time, without stopping the raid, busting them apart, and
comparing them?

These are production machines that are acting funny, and I just wonder
if the cause is out of syncness between the two raid 1 drives. The system
CLAIMS they are in sync, I can always fail and reinsert a drive, but
I would like a way to test the sync without doing all that. Seems
like there should be a way to just mdadm --verify /dev/md1 or something

Thanks in advance,

Homer 607 227 5465

Good grief, I hope that's not a real phone number.

I'm not using software RAID on any of my hosts, but I think the true test would be failing / removing a drive.

A better approach may be starting a new thread, explaining the exact symptoms of your servers that are "acting funny". It would be a good idea to establish cause and effect rather than to play around with your RAID on a whim.

you can examine the status of the raid array using

cat /proc/mdstatus I think its called. It will show the status of the RAID array, useful for monitoring the re-sync process. If the disks are out of sync then there is further problems going on.

watch cat /proc/mdstatus

Thanks to those that answered.

Issue is I *BELIEVE* I have at one time or another had
two drives that were in sync, made to be out of sync by
individually mounting them, changing one, and then rebooting
as an array. The array did not come up 'dirty', and as an array
it showed the data on the one drive, but not the other. Then when
I took them apart again, the two drives were definitely different.

Thus if there is a hard ware problem that is not writing
data to both drives properly, or there is a crash that interrupts
the process, I have no clue, it is conceivable that the two
drives would be considered clean when not.

Since different data would exist in each drive, and since
I believe reading is random choice between drives, it would be
possible for the OS to read the wrong data once in a while and
collapse.

I just know that code that depends on sync between two files,
to work but never checks the sync with read after write, or some
other mechanism, is just bound to fail someday.

It would be nice to have a verify option, that didn't involve
destroying the backup copy for the duration of the recopy, but simply
read both drives in the back ground and reported any problems.
Seems obvious but then I am generally clueless about about how all
this works.

Homer