Help With Degraded Linux Raid

[Netopia]

Hi guys,

I'm not quite sure what to do. I've got a system with two drives setup in Raid-1. I had some hiccups on the machine the other day and now I've got this situation (see code).

So how do I remirror just one device out of a whole disk? Any ideas?

Joe

 

[Netopia]

No love? 😉

 

[Nothinman]

You could probably remove and hotadd /dev/hda3, I don't really know for sure.

And why does it say device 0 is removed from md1?

 

[Netopia]

Originally posted by: Nothinman

And why does it say device 0 is removed from md1?

Hmmmm.... good question! I had somehow missed that and simply thought it referred to /dev/hda3.

Dunnno!

Joe

 

[EatSpam]

Looks like that partition got knocked out of the array. The drive probably took too long performing an error correction or something. I've had similar situations with my Linux RAID 5 arrays.

You want to hotremove then hotadd /dev/hda3. Check my syntax, but I think its right.

remove: mdadm /dev/md1 -r /dev/hda3
add: mdadm /dev/md1 -a /dev/hda3

With any luck, it'll say successful to both of those commands and the RAID will rebuild. Use the command cat /proc/mdstat to monitor.

 

[EatSpam]

 

[drag]

I had a similar thing happen a couple times. I think it was due to some faulty RAM or something like that because after I replaced that ram it seemed to have gone away. Either that or the drive it was on was overheating (I moved things around after that also to make sure that everything had good airflow)

I simply re-added the partition back to the array and it seemed to have liked it. After a rebuild it ran fine.

Probably be a good idea to keep a good eye on it until you figure out why it happenned. Power failure, hardware failure, software bug, whatever. But you said 'hiccups' so it's likely you already knew what happenned. 🙂

If they are regular IDE devices you can use SMART tools to access the drive's internal diagnostic information and run some tests. For sata devices these smart tools are only aviable for very new versions of the libata drivers and such.

 

[Netopia]

I had built this machine to do antispam/antivirus (http://www.freespamfilter.org) and it was finally set up and totally beta tested and kicking some serious spam butt (~90% of traffic killed as spam with almost zero false positives). I moved the machine into the server room and it didn't want to boot a couple of times... and then it finally booted...very strange. The next day my log watch showed the problem and it was right at the time that the failed boot happened.

🙁

 

[Netopia]

Originally posted by: EatSpam
Looks like that partition got knocked out of the array. The drive probably took too long performing an error correction or something. I've had similar situations with my Linux RAID 5 arrays.

You want to hotremove then hotadd /dev/hda3. Check my syntax, but I think its right.

remove: mdadm /dev/md1 -r /dev/hda3
add: mdadm /dev/md1 -a /dev/hda3

With any luck, it'll say successful to both of those commands and the RAID will rebuild. Use the command cat /proc/mdstat to monitor.

I'll try that! THAT is what I was looking for! I'll have to read the man pages on mdadm and get familiar with it. Sometimes this stuff is SOOOOO frustrating!!!! Today, I found a ton of pages about how to do RAID in Linux and almost ALL of them refered to Raidtools2. I found instructions for installing Raidtools2 via apt-get, so I went out and found out how to install apt-get for FC4. Then I tried to install Raidtools2, but it failed. HOURS later, I find a page that says that Raidtools2 is no longer being developed and that I should use mdadm! ARGHGHGHGH... mdadm is installed by default!

I'm beginning to both love and hate linux. I love it for all the reasons that everyone already knows... I probably hate it for the reasons everyone knows too... but there is so much mis/old information out there that you can waste HOURS chasing wild geese!

But EatSpam... thanks for the direction pointing. If your username indicates a loathing of spam, why not come on over to http://www.freespamfilter.org and take a look. You might learn something, but more importantly, I'm guessing you may be able to help others afflicted with the evil pork product!

Joe