dwielgosz
asked on
Any ideas on what this error means?
We have ES3.0 (Redhat) and it runs very well for the most part. It's used to host our database accounting application. I ran into some problems this AM (unusual). First of all, I couldn't get the tape to eject from the drive. Finally I decided to reboot the server and get it out then. While shutting down, I lost the X window and had a black screen with line after line of this:
"ext3-fs error (device sd(8, 5)) in start-transaction: Journal has aborted"
Like I said, line after line after line of that. I couldn't get the machine to respond so I tried getting a terminal window up with the F keys. When a terminal would open, after a moment, I would get:
"journal committ I/O error"
Finally it rebooted
Any ideas on what the two errors are caused by, or what they are referring to?
"ext3-fs error (device sd(8, 5)) in start-transaction: Journal has aborted"
Like I said, line after line after line of that. I couldn't get the machine to respond so I tried getting a terminal window up with the F keys. When a terminal would open, after a moment, I would get:
"journal committ I/O error"
Finally it rebooted
Any ideas on what the two errors are caused by, or what they are referring to?
rindi
Ext3 is as opposed to the older ext2 a journaling filesystem. Journaling filesystems keep journals of their transactions inside reserved space. This allows the filesystem to be repaired in case of the disk being disconnected or turned of without properly shutting the system down by replaying these journals to the original filesystem. Something has now probably happenned to your filesystem which the journal can't restore, and that seems to be on sd8, partition 5. Maybe that device is bad.
ASKER
How can I check that in Redhat ES? The integrety of that partition?
While the partition isn't mounted run fsck.ext3 (if it is a system partition you might have to be in single user mode). Without options you should get the correct syntax.
ASKER
I'm really green at Linux. Could you tell me how to determine if it's a system partition? And by "single user mode" do you mean open an F1 terminal and log off as root?
When you boot up there should be a boot menu, and if there is something similar to "recovery" or "disaster" or similar, boot into that mode. Then you are in single user mode. Then you don't need to worry whether this is a system partition or not (System Partition where files needed by the OS are on).
In order to work out which partition is which, type:
mount
after a normal startup, and you'll get a list of your mounted partitions. This also gives details of whick partitions are mounted where:
man mount
for more info.
When booting into 'recovery' mode, you often do this from your 'boot CD'. In which case, the boot process will get as far as the prompt:
linux:
At this prompt, enter either:
1
or
single
and press enter.
You should then boot into safe mode (or single user mode).
Alternatively, you can boot into single mode via your normal 'grub' bootloader. You will need to edit the command similar to:
kernel /vmlinuz-2.4.21-15.EL ro root=LABEL=/
and change it to:
kernel /vmlinuz-2.4.21-15.EL ro root=LABEL=/ single before booting, and you'll boot into single mode.
fsck.ext3 is in the /sbin directory, so you may need to enter the path in full to get it to work:
/sbin/fsck.ext3
This assumes that the /sbin partition is mounted. If the /sbin directory is on sd8, then you ought to carry out the process using your recovery/install CD. This is because, partition repairs should only be attempted when they are not mounted.
man msck.ext3 in order to find out exactly which options you want to use, although running it without any 'switches', (- switches) should be fine for the first run.
HTH:)
mount
after a normal startup, and you'll get a list of your mounted partitions. This also gives details of whick partitions are mounted where:
man mount
for more info.
When booting into 'recovery' mode, you often do this from your 'boot CD'. In which case, the boot process will get as far as the prompt:
linux:
At this prompt, enter either:
1
or
single
and press enter.
You should then boot into safe mode (or single user mode).
Alternatively, you can boot into single mode via your normal 'grub' bootloader. You will need to edit the command similar to:
kernel /vmlinuz-2.4.21-15.EL ro root=LABEL=/
and change it to:
kernel /vmlinuz-2.4.21-15.EL ro root=LABEL=/ single before booting, and you'll boot into single mode.
fsck.ext3 is in the /sbin directory, so you may need to enter the path in full to get it to work:
/sbin/fsck.ext3
This assumes that the /sbin partition is mounted. If the /sbin directory is on sd8, then you ought to carry out the process using your recovery/install CD. This is because, partition repairs should only be attempted when they are not mounted.
man msck.ext3 in order to find out exactly which options you want to use, although running it without any 'switches', (- switches) should be fine for the first run.
HTH:)
ASKER
This is what "mount" returns:
[root@REDHAT root]# mount
/dev/sda5 on / type ext3 (rw)
none on /proc type proc (rw)
none on /dev/pts type devpts (rw,gid=5,mode=620)
usbdevfs on /proc/bus/usb type usbdevfs (rw)
/dev/sda3 on /appl type ext3 (rw)
/dev/sda1 on /boot type ext3 (rw)
none on /dev/shm type tmpfs (rw)
/dev/sda2 on /u type ext3 (rw)
[root@REDHAT root]#
"REDHAT" is the name of the server btw.
[root@REDHAT root]# mount
/dev/sda5 on / type ext3 (rw)
none on /proc type proc (rw)
none on /dev/pts type devpts (rw,gid=5,mode=620)
usbdevfs on /proc/bus/usb type usbdevfs (rw)
/dev/sda3 on /appl type ext3 (rw)
/dev/sda1 on /boot type ext3 (rw)
none on /dev/shm type tmpfs (rw)
/dev/sda2 on /u type ext3 (rw)
[root@REDHAT root]#
"REDHAT" is the name of the server btw.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
"You should be able to narow down the list and pinpoint to failing or malfunctioning device
( I think it was your SCSI tape backup unit )
"journal committ I/O error" basically translates to whether your device is malfunctioning, it isn't mounted , files system is corrupt or simply
device is unavailable for kernel to execute scheduled task ( journaling fs every 5 seconds )"
This really hit it right on target. It is a failing tape drive. The above explanation made it crystal clear for me. Thanks so much for the information and sharing of your knowledge.
Dave
( I think it was your SCSI tape backup unit )
"journal committ I/O error" basically translates to whether your device is malfunctioning, it isn't mounted , files system is corrupt or simply
device is unavailable for kernel to execute scheduled task ( journaling fs every 5 seconds )"
This really hit it right on target. It is a failing tape drive. The above explanation made it crystal clear for me. Thanks so much for the information and sharing of your knowledge.
Dave
Great!
nedvis
nedvis