LinuxQuestions.org
Visit Jeremy's Blog.
Go Back   LinuxQuestions.org > Forums > Linux Forums > Linux - General
User Name
Password
Linux - General This Linux forum is for general Linux questions and discussion.
If it is Linux Related and doesn't seem to fit in any other forum then this is the place.

Notices


Reply
  Search this Thread
Old 01-07-2006, 03:55 PM   #1
mazzo
Member
 
Registered: Jun 2003
Location: Thames Valley, UK
Distribution: RedHat from 5 -9, Fedora, Ubuntu, Centos 3 - 7, Puppy Linux
Posts: 140

Rep: Reputation: 15
Is this a drive problem?


Hi

I am getting an error which occurs at the same time every day.

Message logs give:

Dec 4 04:02:34 Polata kernel: Directory sread (sector 0x13) failed
Dec 4 04:02:34 Polata kernel: attempt to access beyond end of device
Dec 4 04:02:34 Polata kernel: 02:00: rw=0, want=10, limit=4


Bizzarely it occurs every day at the same time and each incident lasts about only about a second or two - but has multiple entries in the log.

Is this a drive error (it's been doing it for months now)?

What is causing it as I am not aware of anything in cron that runs at 4:02? How can I find out and what can I do to stop/fix it without killing my data?

It is a Red Hat 9 server 2.4.20-43.9. kernel

Thanks in advance.
 
Old 01-07-2006, 04:12 PM   #2
gilead
Senior Member
 
Registered: Dec 2005
Location: Brisbane, Australia
Distribution: Slackware64 14.0
Posts: 4,138

Rep: Reputation: 168Reputation: 168
Weird - it sounds like a cron job. Can you run the following:

Code:
grep '[[:digit:]][[:digit:]]* 4' /var/spool/cron/crontabs/*
That will return anything in your crontabs set to run between 4:00am and 4:59am.
 
Old 01-08-2006, 12:51 AM   #3
saturndude
Member
 
Registered: Mar 2005
Location: across the river from Louisville KY
Distribution: Mandriva 2010.2 (64-bit)
Posts: 57

Rep: Reputation: 15
Beyond end of device

Hi,

I ran into some "beyond end of device" errors a few years back. How old is your drive, and what OS was used to set it up originally? There are several different hard drive geometries out there. In the "old days", two major geometries were used by BIOS makers:

256 heads * 63 sectors/track * 512 bytes/sector = 8,257,536 bytes per cylinder (LBA)
(heads numbered 0 through 255)

240 heads * 63 sectors/track * 512 bytes/sector = 7,741,440 bytes per cylinder (LBA)
(heads numbered 0 through 239, or may use 16 heads and "LARGE" selection in BIOS)

That second one was used in the past by Intel chip sets. Now, a third geometry has arisen:

Units = cylinders of 16065 sectors * 512 bytes/sector = 8,225,280 bytes per cylinder

The drive starts with cylinder 0 and traditionally partitions can only end at the end of a cylinder.

The partition table tells how the partitions were set up when the drive was first used. If a partition begins or ends beyond the 8.4 Gig point (past cylinder 1023), well, there is no space to write a cylinder number larger than 1023, so another method is used. Another number is recorded, telling how many sectors from the beginning of the drive you must go to reach the start of each partition (Norton Utilities 2001 calls this number BIG TOTAL # OF SECTORS).

If the BIOS and Operating System see the drive differently from what was recorded in the partition table, their calculation of the "BIG TOTAL # OF SECTORS" will differ from the partition table numbers, and partitions may appear to overlap each other or extend past the end of the drive (bad things will happen).


The time (4:02 AM) suggests some kind of system maintenance-type task.

If you can find the reason the cron job thinks your disk is different than it is, you are a better person than I am. I would leave this alone if there is no harm until the next computer purchase or Linux upgrade, but only you can judge that.


Let us know how it turns out!

saturndude

Last edited by saturndude; 01-08-2006 at 12:58 AM.
 
Old 01-08-2006, 08:50 PM   #4
saturndude
Member
 
Registered: Mar 2005
Location: across the river from Louisville KY
Distribution: Mandriva 2010.2 (64-bit)
Posts: 57

Rep: Reputation: 15
Update

Sorry about the un-original subject line.

I noticed you are using a 2.4 kernel. In the 2.6 kernel, detecting the hard drive geometry was taken out of the kernel and put into "userspace".

I don't know what that means, but when I upgraded to Mandrake 10.0 (2.6 kernel) it detected my partitions incorrectly and I allowed it to alter the partition table. This caused problems with partitions not ending on a cylinder boundary, "beyond end of drive", etc.

I suspect Fedora Core uses the 2.6 kernel. If FC is on the same machine as the polata kernel, and FC modified the partition table, FC could mess things up for every OS on that drive.

My US $0.02.

saturndude
 
Old 01-09-2006, 04:17 AM   #5
mazzo
Member
 
Registered: Jun 2003
Location: Thames Valley, UK
Distribution: RedHat from 5 -9, Fedora, Ubuntu, Centos 3 - 7, Puppy Linux
Posts: 140

Original Poster
Rep: Reputation: 15
Thank you to both.

It was a fresh install onto the drive. I don't remember any messages about geometry. I put this server together in early 2003 and it is in constant use. It stays on, rarely gets a reboot (unless a new kernel is out) and is very reliable.

So far it has not caused any file problems so I am tempted to let sleeping dogs lie (although this sleeping dog is twitching!)

I will try to find out what is going on in cron. Couldn't find anything when I looked before, but I will use gilead's idea to check.

Will report back on this when I can. Not in the office for a couple of days so it won't be instantaneous.

Thank you for your help so far.
 
Old 01-11-2006, 05:44 AM   #6
mazzo
Member
 
Registered: Jun 2003
Location: Thames Valley, UK
Distribution: RedHat from 5 -9, Fedora, Ubuntu, Centos 3 - 7, Puppy Linux
Posts: 140

Original Poster
Rep: Reputation: 15
The grep produces nothing! I can't find anything that is scheduled to run at 4am!

Odd. The command wasn't quite right, however, so I adjusted it. The path is /var/spool/cron and the only two files are a tmp.1234 and root.

Still a log-full of errors from the last couple of days. (I don't think I was expecting it to just go away!!!! )
 
Old 01-11-2006, 12:45 PM   #7
gilead
Senior Member
 
Registered: Dec 2005
Location: Brisbane, Australia
Distribution: Slackware64 14.0
Posts: 4,138

Rep: Reputation: 168Reputation: 168
Just out of curiousity what is in the file called root?

The only thing Googling returned was the possibility that you had a removable device mounted at some stage and removed it without mounting and now something like slocate is trying to scan that device. For example https://www.redhat.com/archives/redh.../msg00990.html or https://www.redhat.com/archives/valh.../msg00354.html

Does `mount` or `cat /etc/mtab` show any removable drives mounted that you didn't expect?
 
Old 01-12-2006, 06:58 AM   #8
mazzo
Member
 
Registered: Jun 2003
Location: Thames Valley, UK
Distribution: RedHat from 5 -9, Fedora, Ubuntu, Centos 3 - 7, Puppy Linux
Posts: 140

Original Poster
Rep: Reputation: 15
Thanks Steve

No devices mounted that shouldn't be. The "root" file contains my fetchmail cron (ie the same contents as you would see if you did crontab -e)

Wierd one. Just hope it doesn't go phut when I least need it!!

Maz
 
  


Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off



Similar Threads
Thread Thread Starter Forum Replies Last Post
Drive problem Draiocht Slackware 36 09-29-2005 07:12 PM
The drive problem Gins Solaris / OpenSolaris 3 09-13-2005 07:48 PM
Slave drive problem Damien_III Linux - Hardware 2 04-08-2005 01:57 AM
problem seeing cd drive class_struggle Linux - Newbie 3 08-18-2004 01:23 PM
zip drive problem (or more specifically, imm problem) CartersAdvocate Slackware 3 02-06-2004 01:01 PM

LinuxQuestions.org > Forums > Linux Forums > Linux - General

All times are GMT -5. The time now is 04:22 AM.

Main Menu
Advertisement
My LQ
Write for LQ
LinuxQuestions.org is looking for people interested in writing Editorials, Articles, Reviews, and more. If you'd like to contribute content, let us know.
Main Menu
Syndicate
RSS1  Latest Threads
RSS1  LQ News
Twitter: @linuxquestions
Open Source Consulting | Domain Registration