[CLUE-Tech] RAID 1 on Linux

[Nate Duehr]

Carl Schelin wrote:

>[Heavy snippage]
>
>  
>
[Ditto.]  (GRIN)

>--- Nate Duehr <nate at natetech.com> wrote:
>
>  
>
>>On Oct 19, 2004, at 7:41 PM, Carl Schelin wrote:
>>    
>>
>>I finally figured out most of this from an article
>>in Sysadmin magazine 
>>about it.  Unfortunately I don't think this
>>particular article is 
>>available online anywhere.
>>    
>>
>
>Well there's the problem. After reading a couple of
>issues I felt it was useless even as toilet paper, due
>to the clay content :-/  I wouldn't have considered
>looking there.
>
>  
>
They do each magazine to a different "theme" each month - after watching 
for a few years I only really find the "security" and "storage" issues 
interesting, most years.

>Being horrible with names, I'm sure I've seen you on
>the co-sage list. Are you the same guy who came in
>with Wendy at the recent meeting (just trying to lock
>the name in with the face)?
>  
>
No, I've actually only made it to one CLUE meeting this year, and my 
schedule recently changed so I can't make it to anything in the 
evenings, BLUG, CLUE... oh well.  (My work schedule is now noon-9PM.  
Kinda weird, but I'm a night-owl, so I'm okay with it for a while.)

>>Here's the rub though -- software RAID1 on 2.4
>>kernels from hard 
>>testing I read on some of the Debian mailing lists
>>from folks like 
>>Russel Coker who wrote bonnie++, shows that there's
>>NO intelligence 
>>about read performance in a Linux Kernel software
>>RAID-1.  It *always* 
>>reads from a single disk, and writes to both.
>>    
>>
>
>I'd like more info on this. In particular, was the
>testing on a specific account? In other words, could
>it be session related so that the system uses a
>particular disk for a process/session but it does
>switch up. Not the best solution, just wondering.
>  
>
To be honest, I don't know.  I'd say Russell would remember or have a 
document published somewhere, though.  He's still very active on 
debian-isp and other debian lists.

>> It
>>gives you 
>>zero-performance-gain for reads, which a lot of
>>Solaris admins would 
>>expect to see from their much more mature software
>>RAID software. 
>>    
>>
>
>I'm not sure I expected it to be mature. I do expect
>it to follow the definition of RAID though. It's
>possible that I'm reading someone's opinion of how
>cool it works on Solaris rather than how it's supposed
>to work. In any case, I'm not too concerned about
>performance. More with having a decent chance of not
>losing data in the event of a disk failure.
>  
>
Yeah, that's what I wanted to use it for too... then I lost data and had 
to restore from a backup.  Once bitten, twice shy, now.

>Could you have a process monitoring the status of the
>RAID (like mdadm monitor mode) and flag it as failed?
>In this case, it's getting to be less and less likely
>I'd recommend RAID 1 on linux to anyone. If I was
>going with RAID 5 or higher, I'd certainly go with
>SCSI.
>
>  
>
I was going to whip up a script to watch /proc/mdstat but it blew up 
before I got that far.  ;-)

Nate