Skip to contentSkip to conversation

Problem with mke2fs on huge RAID-partition

DISCUSSION Peter H. Rüegg 5 participants 8 messages Feb 8, 2002 – Feb 12, 2002

lore
Share & tools
lore
#0 lore raw

Hi there,

I just build a nice little fileserver consisting of 2 40GB-HD's with some RAID-1 partitions and 6 80 GB-HD's on 3 Promise Ultra 100 TX-Adapters as a RAID-5 partition.

All RAID is done in software, using (at the moment) a Standard RedHat 7.2 Kernel 2.4.7.

My problem is: If I start mke2fs [1] on the device, it writes everything down until "Writing Superblocks...". The system then completly hangs. And yes, I did wait long enough (well, at least I think 15 hours should be enough ;-)

Is there a limitation in the maximum size of a partition (well, 400 GB is not that small...), may it be a (known) problem of mke2fs or the particular Kernel-Version, or does anyone have any suggestions where else to seek?

Thanks in advance

Peter H. Ruegg
Systems-/Networkadministrator eProduction AG

--8<------------------------------------------------------------------------- main(){char*s="O_>>^PQAHBbPQAHBbPOOH^^PAAHBJPAAHBbPA_H>BB";int i,j,k=1,l,m,n; for(j=0;j<7;j++)for(l=0;m=l-6+j,i=m/6,n=j*6+i,k=1<<m%6,l<41-j;l++) putchar(l<6-j?' ':l==40-j?'\n':k&&s[n]&k?'*':' ');}

#1 lore raw
Quoted 8 lines · Peter H. R|egg

My problem is: If I start mke2fs [1] on the device, it writes everything down until "Writing Superblocks...". The system then completly hangs. And yes, I did wait long enough (well, at least I think 15 hours should be enough ;-)

Is there a limitation in the maximum size of a partition (well, 400 GB is not that small...), may it be a (known) problem of mke2fs or the particular Kernel-Version, or does anyone have any suggestions where else to seek?

Well, I know for a fact that people have created such large ext2 filesystems.

Quoted 2 lines

All RAID is done in software, using (at the moment) a Standard RedHat 7.2 Kernel 2.4.7.

The first thing to do would be to update to the latest RH kernel.

Cheers, Andreas
--
Andreas Dilger
http://sourceforge.net/projects/ext2resize/
http://www-mddsp.enel.ucalgary.ca/People/adilger/

#2 lore raw
In reply to Peter H. Rüegg
Quoted 4 lines

My problem is: If I start mke2fs [1] on the device, it writes everything down until "Writing Superblocks...". The system then completly hangs. And yes, I did wait long enough (well, at least I think 15 hours should be enough ;-)

More than enough

Quoted 3 lines

Is there a limitation in the maximum size of a partition (well, 400 GB is not that small...), may it be a (known) problem of mke2fs or the particular Kernel-Version, or does anyone have any suggestions where else to seek?

The limit is about 1Tb currently. You are hitting something else, perhaps a driver or VM problem ?

#3 lore raw
In reply to Peter H. Rüegg

Alan Cox <[email protected]> :
[...]

Quoted 2 lines

The limit is about 1Tb currently. You are hitting something else, perhaps a driver or VM problem ?

Promise driver + 2.4.17, see:
Message-ID: <[email protected]>
Message-ID: <[email protected]>

| Intel | Promise
-----------------+-------+--------

raid1 creation   |  fast |  fast
dd of=filesystem |  fast |  slow (*)

mkfs doesn't behaves too badly but it did when I first tried to raid1 the whole disks.

Raid1 is software only.
As soon as a filesystem on the promise adapter comes into play, writes maxes out at 2,5Mo/s. The previous machine (old PA2012 motherboard) with 8 times less memory was able to stand 4~5Mo/s with vanilla broken kernel. Now it's running 2.4.18-pre3-ac2 but the behavior is the same with vanilla pre, vanilla + akpm ll, +ide patches. Feel free to ask if you want a test on a specific version. I have dedicated a partition on each disk for testing.

Signature · 1 line
Ueimor
#4 lore raw
Quoted 7 lines · Francois Romieu

Raid1 is software only.
As soon as a filesystem on the promise adapter comes into play, writes maxes out at 2,5Mo/s. The previous machine (old PA2012 motherboard) with 8 times less memory was able to stand 4~5Mo/s with vanilla broken kernel. Now it's running 2.4.18-pre3-ac2 but the behavior is the same with vanilla pre, vanilla + akpm ll, +ide patches. Feel free to ask if you want a test on a specific version. I have dedicated a partition on each disk for testing.

I don't have a fix for this directly, but the slow speed can be fixed on many Promise controllers with hdparm. Testing the transfer rate of some of my mature (ATA/66) drives, I find that raising the transfer rate from 3MB to 14MB is often possible after setting the options.

NOTE: wrong options will hose your data! WHich is why I don't tell you what to use, just look at -m -c (I use 3), -d and -X34. Again, it may bite you, have backups.

Signature · 3 lines
bill davidsen <[email protected]>
  CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.
#5 lore raw
In reply to Francois Romieu

Bill Davidsen <[email protected]> :
[hdparm]

Quoted 3 lines

NOTE: wrong options will hose your data! WHich is why I don't tell you what to use, just look at -m -c (I use 3), -d and -X34. Again, it may bite you, have backups.

The kernel did itself the job through the "autotune" option of ide. /proc/ide/{hda/hdg}/settings differ only in:
@@ -11,7 +11,7 @@

 failures                0               0               65535           rw
 file_readahead          124             0               16384           rw
 ide_scsi                0               0               1               rw
-init_speed              69              0               69              rw
+init_speed              12              0               69              rw
 io_32bit                3               0               3               rw
 keepsettings            0               0               1               rw
 lun                     0               0               7               rw
@@ -20,7 +20,7 @@
 multcount               0               0               16              rw
 nice1                   1               0               1               rw
 nowerr                  0               0               1               rw
-number                  0               0               3               rw
+number                  2               0               3               rw
 pio_mode                write-only      0               255             w
 slow                    0               0               1               rw
 unmaskirq               1               0               1               rw

It can be fast: it does during raid rebuild.
May be the machine simply dislikes me.

Signature · 1 line
Ueimor
#6 lore raw
Quoted 11 lines · Francois Romieu

Bill Davidsen <[email protected]> :
[hdparm]

Quoted 3 lines

NOTE: wrong options will hose your data! WHich is why I don't tell you what to use, just look at -m -c (I use 3), -d and -X34. Again, it may bite you, have backups.

The kernel did itself the job through the "autotune" option of ide. /proc/ide/{hda/hdg}/settings differ only in:

It can be fast: it does during raid rebuild.
May be the machine simply dislikes me.

What's your stripe size? I have that "this works for me" feeling, although I'd like to know why the drives didn't autotune just the same way, and that might tell someone what's up.

Signature · 3 lines
bill davidsen <[email protected]>
  CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.
#7 lore raw
In reply to Francois Romieu

Bill Davidsen <[email protected]> :
[...]

Quoted 1 line

What's your stripe size? I have that "this works for me" feeling, although

raiddev                 /dev/md20
raid-level              1
nr-raid-disks           2
chunk-size              32k
persistent-superblock   1
nr-spare-disks          0
        device          /dev/hde3
        raid-disk       0
        device          /dev/hdg3
        raid-disk       1

/dev/md10 is built on hda3, hdc3 the same way.

Quoted 2 lines

I'd like to know why the drives didn't autotune just the same way, and that might tell someone what's up.

I'd like too. Datasheet anyone ?

Signature · 1 line
Ueimor