SATA PCI cards and FreeBSD

I have an old computer running FreeBSD headless as a file server. It's a PIII system with an Intel mobo. I believe 384MB RAM? Somewhere around there.

I need to add more storage, and would like the speeds of SATA. Looking at PCI 33/66mhz cards, but having issues finding one that is known to support freebsd.

I do not need to boot from these drives, just add more storage space.

Does anyone know of any cards that do support this? I found a bunch of options on amazon under $20 bucks, but cant find whether they will work or not.

Trying to save a bunch of time shipping stuff, testing them, shipping back, repeat...
 
Odd, the documentation seems to be all about supported SATA RAID controllers with no mention of normal SATA chipsets.

Regardless, I use this card. Anything with the same VIA 6421 chipset should work just as well, and there are countless other supported chipsets I can't name.
 
jnr said:
Odd, the documentation seems to be all about supported SATA RAID controllers with no mention of normal SATA chipsets.

Regardless, I use this card. Anything with the same VIA 6421 chipset should work just as well, and there are countless other supported chipsets I can't name.

Interesting. That was one of the cards I was looking at and hoping someone would say it worked for them.

Awesome. Thanks man!
 
I've a tx4, and while it works well, sometimes it
(writing, say, rsync ... to the sata2)
panics and makes the target filesystem non-fsck'able.
So I wonder if the above card (sata1 rather than sata2)
will slow things down enough to not crash it as
often, every other year would suit me fine. OTOH I
should check the sata2 disk to see if there is a
jumper to do the same result...
(no, it is a downloadable executable),
( _8 stable here BTW)

That would make the
tx4 suitable for gm0, zfs, gjournal, etc maybe (unless
their characteristics make recovering from such
a crash just a matter of reslivering or something...
 
I had a SiS chipset on a sweex PCI Sata controller and it caused my system to panic (very) often. This happend mainly when there were a lot disk writes, like copying a few GB to one of the two attached disks, or just with a fsck_ufs background check.

Today I replaced it with a promise tx4 controller (http://www.newegg.com/Product/Product.aspx?Item=N82E16816102065) and so far it seems to be working fine. Found a lot of positive information on the mailinglists. But the post above worries me a bit.

The promise controller is not the cheapest, as you can see on the link, but if makes the system reliable again it's money well spend imo!
 
Tried to copy a dump file (about 1 1/2 G) and
it failed at somewhere about 500M.
(cannot determine filesystem type, the FS was still there
but cannot be fsck'd) after panic.

Need maybe
to find a "slower cp", from the error messages
one that copies "slowly", taking thrice as
long ?? without buffers? to not overload the
controller chipset ?? dd somehow? Not investigated
it much so far...
 
Seems like a Solution... (see last few lines)

jb_fvwm2 said:
Tried to copy a dump file (about 1 1/2 G) and
it failed at somewhere about 500M.
(cannot determine filesystem type, the FS was still there
but cannot be fsck'd) after panic.

Need maybe
to find a "slower cp", from the error messages
one that copies "slowly", taking thrice as
long ?? without buffers? to not overload the
controller chipset ?? dd somehow? Not investigated
it much so far...

Found *maybe* a useful solution to that (or any other
fast-io crashing sata pci controllers maybe)...
..........
use --bw-limit=1000 (#first usage, beta)
the primary disk is sending at .99/mb/sec
and the sata-on-pci is receiving at
.10 mb/sec. with #rsync
Code:
#source directory#
rsync -vaH --partial --stats --numeric-ids \
--inplace --archive --compress --hard-links \
--bwlimit=1000 . /dest
Don't know if that is the optimum CLI but seems
usable (maybe even the parameter with /rsnapshot/ )
at least in the first test case here...:e

Code:
 --one-file-system  #copied fat32 anyway...
Ran out of time configuring gmirror. Configured gjournal per
simple man page example. Could not wipe gmirror with fdisk,
used BootIt.
Rsync passed this test (backup of /) easily.
(no Bsdlabel on the gjournaled partition yet, so this post is BETA/newbie in that
regard). Yes, there was/is (just a two-liner).

update...
Sata on pci tx4 now "usable", backed up
usr / var tmp (part of an hour, 3, usr -- 3 or 4
hours) without problem AT LAST. Onto
journaled partition ( I can postpone
gmirror studying, ignore crashes/fsck's
because of backup probably, use the tx4
reliably, schedule the backups etc etc...
hopefully this "beta" post can remain
informative )
 
Back
Top