I'm always nervous about pulling the plug on a system... could cause file
corruption in that act alone if a write is in progress.

My local friendly RadioShack has a special on SanDisk Ultra II 2GB CF cards,
$19.99.   2GB seams a bit overkill but so what, I got one and am about to
start the process of loading it up.  I'll also redirect syslog to ram (or
maybe another server I have here) and I'm going to take a look inside the
asterisk gui and see if I can get it to use a temp location other than its
source directory... /var/spool/asterisk/tmp would seam appropriate.
Crossing fingers for a more stable system!

Thanks,
David

On Thu, Jan 1, 2009 at 11:50 PM, Darrick Hartman
<dhart...@djhsolutions.com>wrote:

> David,
>
> I had this happen with one box over the summer.  There were some other
> (let's call them environmental) issues that caused the on-site personnel to
> think they could solve the problem by pulling the plug on the system.  Most
> file system problems should be resolved by fsck automatically on reboot.  If
> the automatic check doesn't resolve the problems (like it's not able to do
> in your case) the file system must be severely damaged.
>
> You have TOTAL control over both CDR and syslogs.  If you look at the
> /etc/init.d/asterisk file, the cdr files are only stored on the keydisk or
> unionfs partition if /mnt/kd/cdr-csv or /mnt/kd/cdr-custom exist.  If
> neither exist, all data is written to a tmpfs ramdisk.
>
> Similarly, if PERSISTLOG is not set to yes, then all syslog data is sent to
> a ramdisk.
>
> I have no control of the behavior of the asterisk-gui.  We only included it
> since several people have requested it's inclusion.  It has improved.
>
> Based on your statement that you're using a cheap compact flash card, I
> would say that a majority of your problems are the card itself and not
> unionfs or any other mechanism.  Some compact flash cards are just plain
> JUNK!  I've been happy with Transcend and SanDisk.  Soekris recommends ONLY
> using SanDisk.  Some of the cheaper CF cards don't properly implement all of
> the functions that these other cards do.  Using one of these other cards is
> very risky!  Even the Transcend cards don't properly support the
> cable-select feature required to have the Soekris net5501 boards call the CF
> card as /dev/hda.  (most of the Transcend cards show up as /dev/hdb).
>
> I have several systems running 0.6 builds (0.6.2 and later) on net5501 or
> VIA targets with none of these same problems.  I'm using systems that have
> as little as 128MB of ram (my home office system) and a new 8 port Digium
> analog card up to a net5501 with 8 analog lines (plus a few others with SIP
> only).
>
> If there is a problem, we certainly want to determine the cause.  Please
> switch to a SanDisk card and see if the problems persist.  If you can't get
> a SanDisk card.  See if you can get a Transcend or a Silicon Systems card.
>  Wim has some of these at www.kd85.com  Silicon Systems has a license from
> SanDisk for some of their technology for wear leveling and other features.
>
> http://www.siliconsystems.com/pdf/SiliconDriveCF.pdf
>
> Specifically look at the wear leveling sections of that document.
>
> Darrick
>
>
> David Kerr wrote:
>
>> I've been having major stability issues over the last week with my alix
>> box and 0.6 build.  I have tracked it down to disk corruption taking place
>> on the CF card in the asturw partition. Things seem to be exasperated by
>> constant updates to the flash card from CDR and syslogs, etc that we've been
>> discussing in another thread.  Here is an example of today's heartache...
>>
>> On booting...
>>
>> Configuring for unionfs...
>> Checking asturw filesystem
>> ext2fs_check_if_mount: No such file or directory while determining whether
>> /dev/
>> hda2 is mounted.
>>
>>
>> ASTURW: UNEXPECTED INCONSISTENCY; RUN fsck MANUALLY.
>>        (i.e., without -a or -p options)
>> Fsck detected errors on /dev/hda2 (4)
>> /bin/sh: can't access tty; job control turned off
>> #
>>
>> Then on removing the CF card and attaching to my CentOS box to diagnose...
>>
>> [r...@localhost ~]# fsck /dev/sdb2
>> fsck 1.39 (29-May-2006)
>> e2fsck 1.39 (29-May-2006)
>> ASTURW contains a file system with errors, check forced.
>> Pass 1: Checking inodes, blocks, and sizes
>> Extended attribute block 131557 has reference count 35, should be 33.
>>  Fix<y>? yes
>>
>> Pass 2: Checking directory structure
>> Entry 'sysinfo_output71.html' in /stat/var/lib/asterisk/static-http/config
>> (61833) has deleted/unused inode 61838.  Clear<y>? yes
>>
>> Pass 3: Checking directory connectivity
>> Pass 4: Checking reference counts
>> Pass 5: Checking group summary information
>> Block bitmap differences:  -149504
>> Fix<y>? yes
>>
>> Free blocks count wrong for group #4 (32269, counted=32270).
>> Fix<y>? yes
>>
>> Free blocks count wrong (211504, counted=211505).
>> Fix<y>? yes
>>
>> Inode bitmap differences:  -61838
>> Fix<y>? yes
>>
>> Free inodes count wrong for group #4 (15442, counted=15443).
>> Fix<y>? yes
>>
>> Free inodes count wrong (107849, counted=107850).
>> Fix<y>? yes
>>
>>
>> ASTURW: ***** FILE SYSTEM WAS MODIFIED *****
>> ASTURW: 342/108192 files (0.6% non-contiguous), 4711/216216 blocks
>> [r...@localhost ~]#
>>
>> After running fsck and placing the CF card back into the alix box, I boot
>> again OK and everything seems fine. But as the system has locked up on me
>> (no calls, no internet access, cannot ssh in) a few times recently I'm
>> getting worried.
>>
>> All I can think of is that my CF card is going bad. It's a cheap Adata 1GB
>> and I have two of them both of which are exhibiting the problem. What would
>> be a more robust card?
>>
>> Thanks
>> David
>>
>
>
>
------------------------------------------------------------------------------
_______________________________________________
Astlinux-users mailing list
Astlinux-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/astlinux-users

Donations to support AstLinux are graciously accepted via PayPal to 
pay...@krisk.org.

Reply via email to