EXT3-fs error Journal has aborted

Guten Morgen,

habe diese Meldung nach einem Absturz in der /var/log/messages gefunden:


Da es sich hier um einen Hardware-Raid-Verbund handelt wundert mich diese Meldung schon.

Hat jemand eine Idee für die Ursache und eine Lösung?

Besten Dank vorne weg...

Hast du schon mal ein fsck.ext3 auf der Partition gemacht?

kann ich erst heute nacht machen - z.Zt. ist der Server im Echtbetrieb.
Melde dann aber!!

Filesystem goes read-only without the underlying storage reporting errors

This document (3735930) is provided subject to the disclaimer at the end of this document.
environment
Novell SUSE Linux Enterprise Server 10
Novell SUSE Linux Enterprise Desktop 10
Novell SUSE Linux Enterprise Server 9
Novell Open Enterprise Server (Linux based)
Novell Linux Desktop 9

ext3, reiserfs or udf filesystem


situation

* Filesystem (ext3, reiserfs or udf) is remounted read-only by the kernel.
* No errors are reported from the storage layer underlying the affected filesystem.



resolution
Novell SUSE Linux Enterprise Server 10
Novell SUSE Linux Enterprise Desktop 10

This issue has been resolved for the ext3 filesystem as of kernel update 2.6.16.27-0.9, released Feb 23, 2007. For the reiserfs filesystem, this issue has been reported to Engineering and is expected to be addressed in the next maintenance update of the kernel packages.

Novell SUSE Linux Enterprise Server 9
Novell Open Enterprise Server (Linux based)
Novell Linux Desktop 9

This issue is resolved for ext3, reiserfs and udf as of kernel 2.6.5-7.286.

status
Reported to Engineering
additional notes
When the filesystem code detects the occurance of an error from which recovery without an fsck is not possible, it changes state to read-only to prevent further corruption.

Normally, this behavior occurs when the underlying storage layer (like SCSI) reported an error when the filesystem code was trying to write out vital data. Errors or warnings reported by the underlying storage layer will preceed the messages from the filesystem layer and the issue can only be addressed by fixing the hardware and/or configuration problem with the storage layer.

With a system under heavy load, a bug in the read-ahead optimisation part of the ext3, reiserfs and udf code could be triggered in which it mistook a temporary error condition from the storage layer ("please retry again later") for an unrecoverable error. In this case, only the filesystem layer reports errors; these errors are not preceded by errors or warnings reported by the underlying storage layer.

In the case of ext3, messages similar to the following are logged:
ERROR: EXT3-fs error (device sda1): ext3_readdir: directory #133120 contains a hole at offset 4096
ERROR: Aborting journal on device sda1.
ERROR: ext3_abort called.
ERROR: EXT3-fs abort (device sda1):
ERROR: ext3_journal_start: Detected aborted journal
ERROR: Remounting filesystem read-only