Filesystems in 7.0 & reliability

Christian Walther cptsalek at gmail.com
Wed Nov 7 11:48:22 PST 2007


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


Hello Noberto,

Norberto Meijome wrote:
> Hi everyone,
> I've been using 7 for a couple of weeks now on my work laptop (kickstarted by cooling issues while in 6.2, which seem to have largely gone in 7).
> 
> I have a 100GB SATA drive in a Thinkpad Z60m with
> 
> CPU: Intel(R) Pentium(R) M processor 2.00GHz (1995.02-MHz 686-class CPU)
>   Origin = "GenuineIntel"  Id = 0x6d8  Stepping = 8
>   Features=0xafe9fbff<FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR,PGE,MCA,CMOV,PAT,CLFLUSH,DTS,ACPI,MMX,FXSR,SSE,SSE2,SS,TM,PBE>
>   Features2=0x180<EST,TM2>
> real memory  = 1609433088 (1534 MB)
> avail memory = 1567780864 (1495 MB)
> 
> I installed 7 on a normal UFS disk, but then migrated to having /usr in a UFS journaled partition , with a 1.5 GB journal on ad0s1h .  I didn't have any issue that I could directly relate to it. I had more hangs than now (i reverted back to plain UFS due to too many lock ups). The lock ups didnt leave any message or error anywhere - it seemed as if the disk subsystem stopped accepting commands (or was waiting on something ...) - anything in memory would work just fine, but as soon as disk access was needed, it 'd stall.
> 

Did you check your harddrive? There are tools available in ports (sorry,
I forgotten how they are called) that can access the drives internal
fault statistics.
Maybe your drive has an error and locks up all of a sudden. Since a
journal leads to more disk activity it would be normal for a hardware
related error to happen more early.

I might be pretty wrong here, but the T60 has an internal movement
sensor that needs some software to turn of the hard drive. Maybe your
laptop has a sensor too, but the logic behind it is implemented in
hardware?

> Something else i also noticed is that, after every crash, I couldn't just reboot and use my computer just fine, as I would have expected - maybe I'm wrong here.
> I had to go into single user mode, and run a fsck /dev/ad0s1f.journal . this takes about 4 minutes. Hardly any errors were ever found  (as opposed to my non-journal partitions, which had files de-referenced ,etc.) So I suppose, in that regards, gjournal worked great.... but is the fsck needed??
> 
> I am also very interested in what zfs has to offer. How reliable is it? I am looking into using it both on my laptop, and as a filesystem for some large storage , possibly.

You might want to search the list archives (especially freebsd-current)
to get some details on ZFS and possible problems regarding it. There
seems to be an issue with ZFS and Samba and/or NFS.
A nice summary is in the archives:

http://lists.freebsd.org/pipermail/freebsd-current/2007-August/076411.html

I don't know what the current status is, thou. And I never suffered from
this error. In fact I'm using a raidz with 4x400GB HDDs. One use is as a
storage for satellite video streams, so there are pretty big files
written to it using NFS.
One thing I really like about ZFS is that it gets rid of the old
partition/slice paradigm. You'll never be angry with yourself again
because you selected the wrong size for your partition/slices.

> 
> thanks for any ideas, comments, pointers :)
> 
> B

HTH
Christian


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (FreeBSD)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFHMh7mND6QF/amlKsRArRMAJ99Siy5hoACgV+DMUF68dtwpkfPDACeLRPq
Lsn2Uk+OBEhh7wgvwJE+FmA=
=ZaeE
-----END PGP SIGNATURE-----


More information about the freebsd-stable mailing list