mkfs taking forever on fiber attached storage

mkfs taking forever on fiber attached storage

Post by jc » Wed, 15 Aug 2001 04:30:29



Greetings,

I have an a5200 storage array with 18gb 10K rpm disks attached to an E420
with a jni pci fiber card, single loop.

I created a radi5 stripe set that is 60gb across 5 physical disks.  When I
issue the command to create the filesystem on the volume /usr/sbin/mkfs -F
ufs /dev/vx/rdsk/datadg/vol01 size it seems to take forever.  I have had
mkfs running for over 2 hours and its still not finished.  I can see the
superblocks being created, but very slowly...... I thought this would rip.
If I do a top I see that rougly 20% of cpu is kernel time and the rest is
iowait.

Anyone have any idea on what tool I can use to see what the problem is, do I
have a problem.?  I found a copuple patch's that seem to be directed towards
vxfs, but I am not uysing that.  I didnt see any hw errors in the messages
file.

Tks

jc

 
 
 

mkfs taking forever on fiber attached storage

Post by Bob Dempse » Wed, 15 Aug 2001 22:42:38



> Greetings,

> I have an a5200 storage array with 18gb 10K rpm disks attached to an E420
> with a jni pci fiber card, single loop.

> I created a radi5 stripe set that is 60gb across 5 physical disks.  When I
> issue the command to create the filesystem on the volume /usr/sbin/mkfs -F
> ufs /dev/vx/rdsk/datadg/vol01 size it seems to take forever.  I have had
> mkfs running for over 2 hours and its still not finished.  I can see the
> superblocks being created, but very slowly...... I thought this would rip.
> If I do a top I see that rougly 20% of cpu is kernel time and the rest is
> iowait.

> Anyone have any idea on what tool I can use to see what the problem is, do I
> have a problem.?  I found a copuple patch's that seem to be directed towards
> vxfs, but I am not uysing that.  I didnt see any hw errors in the messages
> file.

> Tks

> jc

If you think it takes a long time to lay down the filesystem, wait'll you try
to fsck it! In all likelyhood you'll run out of virtual memory to hold all the
inodes (unless this has been fixed in fsck since 2.6). You may want to
investigate the -i option to newfs to reduce the number of inodes. To make best
use of this option, though, you've got to know something about the nature of
the files that are going to be stored on the filesystem.

--
------------------------------------------------------------------------
Bob Dempsey                                     972-550-1133x104(V)
VP Software Development                         972-753-4407(F)

1231 Greenway Drive, Suite 300                  Las Colinas, TX 75038

 
 
 

mkfs taking forever on fiber attached storage

Post by Jean-Francois Land » Thu, 16 Aug 2001 04:09:35



Quote:>Greetings,

>I have an a5200 storage array with 18gb 10K rpm disks attached to an E420
>with a jni pci fiber card, single loop.

>I created a radi5 stripe set that is 60gb across 5 physical disks.  When I
>issue the command to create the filesystem on the volume /usr/sbin/mkfs -F
>ufs /dev/vx/rdsk/datadg/vol01 size it seems to take forever.  I have had
>mkfs running for over 2 hours and its still not finished.  I can see the
>superblocks being created, but very slowly...... I thought this would rip.
>If I do a top I see that rougly 20% of cpu is kernel time and the rest is
>iowait.

Maybe it's just that the array hasn't finished resyncing yet. RAID5
isn't terribly fast for writes when it's up and running, but it's just
plain terrible when it's resyncing :) Use whatever tool you have to
check the status of the array.

Jean-Francois Landry
--
Your eyes are weary from staring at the CRT.  You feel sleepy.  Notice how
restful it is to watch the cursor blink.  Close your eyes.  The opinions
stated above are yours.  You cannot imagine why you ever felt otherwise.
--