|
Hi,
I have a very subtle problem with my HPT370
controller.
Here is my config:
ASUS A7M266
Athlon 1.2Ghz
256MB PC2100 RAM
Video NVIDIA Riva TNT
Network DLink 530TX (not +) 10/100
Raid Highpoint Tech HPT370
The cards are inserted as follows:
Video - AGP
Network - PCI Slot 2
Raid - PCI Slot 3
Interrupts:
Video = 11 (not used)
Network = 10
Raid = 5 (both ide2 and ide3)
My IDE config: (extracts from dmesg)
hda: QUANTUM FIREBALLlct15 30, ATA DISK
drive
hdb: ATAPI 52X CDROM, ATAPI CD/DVD-ROM drive hdc: MAXTOR 4K080H4, ATA DISK drive hdd: MAXTOR 4K080H4, ATA DISK drive hde: Maxtor 4D080H4, ATA DISK drive
hdf: Maxtor 4D080H4, ATA DISK drive hdg: Maxtor 4D080H4, ATA DISK drive hdh: Maxtor 4D080H4, ATA DISK drive hda: 58633344 sectors (30020 MB) w/418KiB Cache,
CHS=62045/15/63, UDMA(66)
hdc: 156301488 sectors (80026 MB) w/2000KiB Cache, CHS=155061/16/63, UDMA(100) hdd: 156301488 sectors (80026 MB) w/2000KiB Cache, CHS=155061/16/63, UDMA(100) hde: 160086528 sectors (81964 MB) w/2048KiB Cache, CHS=158816/16/63, UDMA(100) hdf: 160086528 sectors (81964 MB) w/2048KiB Cache, CHS=158816/16/63, UDMA(100) hdg: 160086528 sectors (81964 MB) w/2048KiB Cache, CHS=158816/16/63, UDMA(100) hdh: 160086528 sectors (81964 MB) w/2048KiB Cache, CHS=158816/16/63, UDMA(100) Found Controller: HPT370 UDMA/ATA100 RAID
Controller
scsi0 : hpt37x2 Vendor: HPT Inc. Model: HPT37x2 RAID 0 Rev: 1.05 Type: Direct-Access ANSI SCSI revision: 02 Attached scsi disk sda at scsi0, channel 0, id 0, lun 0 SCSI device sda: 640344065 512-byte hdwr sectors (327856 MB) sda: sda1 Note: the two "4K" maxtors are setup in stripping
array (2x80GB), using linux's build-in software raid feature. They are working
perfectly fine.
My goal: get the 4 other 80-GB maxtors (4D) to work
in a striping array using the HPT Raid controller. Here is my
story:
First, I have setup everything as usuall. Plugged
in the 4 disks, setup the stripping array from HPT BIOS (1-3-2-4 in stead of
plain 1-2-3-4 disk-in-the-array setup to boost the performance with 64K stripe
size). fdisk'ed the /dev/ataraid/d0. on mkfs.ext3 -m 0 -R stride=16
/dev/ataraid/d0 I get kernel panic.
So I went to HPT's website, got the latest RH7.3
drivers. Did same thing. Now the system freezes (no panic) but in a weird
fashion. When writing inode tables, its stops at about 40-42 (out of 2443) and
just sits there. I can still do Ctrl+F2,F3, etc to change the consoles, however
when I try to login (keybord is working), the systems just sits there and does
nothing after I type my username and password. I guess an io wait for the entire
IDE subsystem or something. I am not a kernel guru :-( But it's better then
kernel panic...
Ok so I went ahead and setup the arrays in a normal
1-2-3-4 disk-in-array mode. This time the fs gets created BUT I get transfer
errors on it. basically I do:
cp /proc/kcore .
cp kcore test
diff kcore test produces Binary files core and test
differ, which of course should not happen. (kcore == 256MB == size of my RAM).
So I think, well, one of my IDE disks is messing up.
I unloaded the hpt37x2 together with scsi_mod and
format every drive separately (fdisk /dev/hdx, setup a primary partition of
maximum size, mkfs.ext3 /dev/hdx, for x=e,f,g,h). Then I did some very heavy
testing of each of the 4 drives (i.e. I created a 1GB file, copied it everywhere
to the disks, did differs, then I copied the 1GIG from hde to hdf and from hdg
to hdh at the same time, did the differs, then hde -> hdg and hdf -> hdh,
did the differs (since ide2 and ide3 share the interrupt, I thought it would be
a good idea to test it that way), etc.. Everthing was FINE. So basically the
board itself, in "plain IDE" setting is working fine. I even did hdparm -t
/dev/hdx for x=e,f,g,h with result of 34MB/sec average for all drives. So
speedwise, the thing is working as well.
So I think, well it must be the RAID controller
itself that is not working.
So I setup 2 RAID-0 arrays (hde - hdf and hdg -
hdh) in stripping mode. Each one is 160GB. I do the tests. AGAIN I GET NO ERRORS
ON COPY (between the two arrays, to my other Linux-software-raid-array and to my
30GB hda). So the thing is working fine. I partition the 4 Maxtors again as a
one huge 320GB stripping array, this time with a stripsize of 256K. (1-2-3-4
since 1-3-2-4 hang on format). the standard test (cp /proc/kcode . cp kcore test
diff kcore test) fails again. Help.
To summarize :
Each of the 4 harddrives is working fine (=> HPT
IDE must be working fine)
Each of the 2 160GB arrays is working fine (=>
HPT RAID must be working fine)
The interrupt sharing for ide2 and ide3 is working
fine (multiple simultaneous copy tests for 2 160GB arrays)
The 1 320GB array produces ERRORS on copy (=>
???????)
When I set it up in 1-3-2-4 disk-in-array mode, the
system hangs (=> even worse ???????? than before)
Unless someone can help me out with anything,
i guess i will just stick with 2x160GB arrays (but that was not the point of the
whole investement). Any help is very appreciated.
I am running RH7.3 with 2.4.18-3 original kernel.
Installed lastest rh package from HPT website version 1.32 (kernel's
hptraid gives me panic every time anyway) Recompiled the provided open source
drivers. Updated the card's BIOS to 2.32.
There is also one more thing that seems strange.
When I format the drives separately, there are 611 node tables (for 80GB
drives). When I format the two 160GB arrays, there are 1222 node tables (pretty
logical: twice the storage, twice the inode tables). BUT, when I do the entire
320GB array, there are only 2443 inode tables (instead of 2444?). I dont know if
it is normal due to the size of the array (ext2 requires more management space
with more size?) or something is up....
Oh and one more thing, I have a 420W AMD-certified
power supply, with 7 fans inside the case (front : 3 - 1 per each hdd cage,
back : 2 and 2 additional fans on the power supply) so I dont think that the
disks are overheating or not getting enough power...(or at least I
hope!)
Any suggestions?
MK
|
