On Wed, 2006-02-22 at 13:37 -0800, jdow wrote:
> From: "Laurence Orchard" <laurence@xxxxxxxxxxxxxxx>
>
> > Hi all
> >
> > when trying to access the web through firefox I get an access denyed
> > error from the proxy server.
> >
> > I have tried to restart squid as listed below and get the following
> > errors:-
> >
> >
> > [root@athlon init.d]# cd /etc/init.d
> > [root@athlon init.d]# ./squid status
> > squid dead but pid file exists
> > squid: ERROR: Could not send signal 0 to process 2828: (3) No such
> > process
> > [root@athlon init.d]# ./squid stop
> > Stopping squid: ./squid: line 84: /var/log/squid/squid.out: Read-only
> > file system
> > [FAILED]
> >
> > [root@athlon init.d]# echo temp > /var/temp
> > bash: /var/temp: Read-only file system
<snip>
> > Feb 17 17:46:40 localhost smartd[2621]: Device: /dev/hdg, 2 Currently
> > unreadable (pending) sectors
>
> My initial suspicion is that you need to hold a decent burial for /dev/hde,
> at least. I'd check that cabling is still OK, that the motherboard settings
> are still OK. And so forth.
>
I am not sure that this IS a hardware problem.
smartd initially started complaining about /dev/hdf, this is a single
filesystem disc of 120Mb that I used only for backup. As I said I remade
the filesystem & checked it, but it found NO errors.
At the moment it is not used and has not been used since it was remade,
but smartd is still complaining of pending sectors.
> Presuming there are no physical or software changes from your last boot
> the above is my best bet. If you changed anything then back off the change.
> For example, boot to an earlier known working kernel. If that doesn't work
> start checking out the individual disks.
the discs are checked each time the system is rebooted with no errors
found. (boy does it take some time!! coffee time!)
>
> Boot to single user mode. Then see what you can do with an fsck on /dev/hde
> through /dev/hdg by hand. If fsck hangs reboot to single user mode and try
> something really basic like "fdisk -l /dev/hde". If that fails try the disk
> on another machine. It MIGHT be the controller being sick. If you get past
> /dev/hde the first time try /dev/hdf and /dev/hdg with this sequence of
> actions. If they all pass fsck you might get away with manually mounting
> all the disks in your system - one at a time - and performing some strategic
> "ls -R" commands to see if the disk can do that much.
been there, done that. when the system reboots everything works fine, no
errors are found at all.
>
> However, it surely looks like one of your disks is dead. If I were to bet
> on it I'd make a tiny wager it was /dev/hde.
>
> > since this last reboot everything seemed to run OK until this morning
I don't think it IS a problem with /dev/hde because /var is only one of
the partitions on this disc, the others /home /dev/hde1
& /windows/D /dev/hde2 have no problem.
> >
> >
> > the last part of /var/log/messages reads:-
> >
> > Feb 22 08:01:01 localhost kernel: SMB connection re-established (-5)
> > Feb 22 08:16:39 localhost smartd[2621]: Device: /dev/hde, 1 Currently
> > unreadable (pending) sectors
> > Feb 22 08:16:49 localhost kernel: hdf: irq timeout: status=0xd0 { Busy }
> > Feb 22 08:16:49 localhost kernel: ide: failed opcode was: 0xb0
> > Feb 22 08:16:49 localhost smartd[2621]: Device: /dev/hdf, not capable of
> > SMART self-check
> > Feb 22 08:17:03 localhost kernel: hdf: irq timeout: status=0xd0 { Busy }
> > Feb 22 08:17:03 localhost kernel: ide: failed opcode was: 0xb0
> > Feb 22 08:17:03 localhost smartd[2621]: Device: /dev/hdg, 2 Currently
> > unreadable (pending) sectors
> > Feb 22 08:46:38 localhost smartd[2621]: Device: /dev/hde, 1 Currently
> > unreadable (pending) sectors
> > Feb 22 08:46:54 localhost kernel: hdf: irq timeout: status=0xd0 { Busy }
> > Feb 22 08:46:54 localhost kernel: ide: failed opcode was: 0xb0
> > Feb 22 08:46:55 localhost smartd[2621]: Device: /dev/hdg, 2 Currently
> > unreadable (pending) sectors
> > Feb 22 09:01:01 localhost mount.smbfs[2049]:
> > [2006^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@^@
> >
> >
> >
> > is it smartd?
>
> That is possible, especially if you can repeat this and smartd was changed
> recently.
>
> > how can I clear the 'unreadable (pending) sectors'?
> >
> > I have done a mkfs on /dev/hdf1 &
> > I have tried running fsck on /dev/hdf1 with the -c -c options but it
> > finds no errors.
> >
> > the partition/file system is empty of data.
>
> It seems /dev/hde is the disk that hits problems first. Test it. "mkfs" is
> not a particularly smart alternative. Although if that is not an NT based
> Windows it might be the right thing to do regardless. Windows 9x is a virus.
> Windows ME is a particularly destructive virus. Windows NT variants are
> pretty good, until you load "somebody else's software". They are generally
> worth keeping if you have good reason, such as "that is where my income is
> based." But none of that solves the basic problem. Do /dev/hde and /dev/hdg
> successfully fsck to the extent you can? (Boot to Windows to chkdsk the
> Windows partitions if you don't simply nuke Windows.) I'd also check the
> boot sector ("fdisk -l /dev/hde") for realistic values. Some Windows malware
> or crudware may have nastied with the values in the boot sector.
>
> {^_^}
I don't run windose on this machine the /windows partitions are
hangovers/backups of data from another machine.
>
