Re: panic after removing usb flash disk



Date: Thu, 16 Aug 2007 10:08:17 -0700
From: Jeremy Chadwick <koitsu@xxxxxxxxxxx>

On Thu, Aug 16, 2007 at 09:31:32AM -0700, Kevin Oberman wrote:
To further complicate things, many of the major contributors to FreeBSD
are only interested in it for its use as a server or embedded OS. This
means that they are willing to commit resources to SMP, which they need,
but not so willing for hot-removal of storage, which is of only slight
value in the server and embedded OS world.

Really? Hmm. This got me thinking: it would benefit Juniper greatly if
they 1) stopped using single disks in their multi-thousand-dollar
routers (try dual disks with RAID 1), 2) stopped using ATA disks and
went with SCSI, and 3) put in a hot-swap backplane of some sort.

Nothing like paying US$20K for a ""enterprise"" product that uses single
ATA disks with no hotswap capability. My point is that it WOULD benefit
some of the major contributors to rank this issue as serious.

I have no idea whether Juniper is a contributor to FreeBSD. Just because
they use it does not mean that they contribute. (Then again, you may
know that they do.)

That said, all of the higher end Juniper boxes really run off Compact
Flash which is ATA. The hard drive is used for backup, distributions,
logging, etc. If properly set up, the system will run fine without a
hard drive. (OK Maybe it limps a bit and you want to get it fixed very
quickly.)

That said, I am quite a ware of the pain caused by hard drive failures
on Junipers. RAID and hot-swap are interesting approaches, but the
ability to pull out a mounted media is probably not too significant. I
hot swap ATA drives in my laptop all of the time. I just umount and use
atacontrol to detach the controller.

In this day of SATA, I'm not sure I see an advantage to SCSI in the
Juniper but there is a need for higher-end drives. They exist in both
3.5 and 2.5 in. form factors and have reliability specs similar to
SCSI. Disk performance is really not an issue for a Juniper router.

I am sure that a lot of people who have no professional interest in
fixing this do have a strong personal interest and I suspect that it
will happen before too long, but complaining about it is not really
going to help as almost every FreeBSD desktop and mobile user has been
bitten by this at one time or another and wants it fixed.

The problem is that as computing changes and the hardware evolves, the
underlying OS design being discussed here has not. There have been many
real-life examples given where one cannot do anything about the
circumstances that induce the panic (USB hub losing power due to a cat
pulling the AC cord, laptops going into S1/S3 sleep mode, or server
admins who need to go to the co-lo and perform some realtime data
copying who simply forgot to umount).

Amen! No argument. I think everyone acknowledges that FreeBSD really
needs to get this fixed. I have tickled this problem in several ways and
it really has a painful bite! (Forgetting to umount a disk on a
server is pretty inexcusable, if human.)
--
R. Kevin Oberman, Network Engineer
Energy Sciences Network (ESnet)
Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab)
E-mail: oberman@xxxxxx Phone: +1 510 486-8634
Key fingerprint:059B 2DDF 031C 9BA3 14A4 EADA 927D EBB3 987B 3751

Attachment: pgpwXTSS8tkrp.pgp
Description: PGP signature



Relevant Pages

  • Re: [SLE] Machine Building - Nightmare Alley - Compatibility and Upgrade
    ... On the other side I've been using Gigabyte boards with SuSE for years no ... graphic cards as the server terminal stays in text mode anyway. ... > department) that I will have to install their SuSe 9.1 distribution from ... Disks have a limited life time... ...
    (SuSE)
  • Re: Unix runs faster, maybe (was: Re: Educating potential VMS users)
    ... That's what I just suggested, Kerry: they're limited by disk I/O, not CPU, and hence CPU utilization *will* be low, even if the disks are working their little tails off. ... A server with very low CPU utilization is a *good* candidate for virtualization, since it's got lots of horsepower left over for other tasks that admins might be reluctant to run on the same OS instance: just hook up some more disks to service the added applicationand let 'er rip. ...
    (comp.os.vms)
  • Re: PowerEdge 1800 Spontaneous Reboots
    ... you're seeing the server crashing? ... I thought it could be the raid controller, but no HDD are attached to ... RAID 1 Mirror array spanning the whole of the 160GB. ... rebuilding the MBR or BOOT record with both disks plugged into the ...
    (microsoft.public.windows.server.sbs)
  • Re: Exchange Disks
    ... I was looking at using a cheaper alternative with 7,200 RPM Server SATA ... Logs raid 1 mirror ... consider single drives for the OS and or Logs instead of mirrors. ... Is there some sort of rule-of-thumb for setting up disks based on a number ...
    (microsoft.public.exchange.setup)