ProblemVERITAS File System (VxFS) 3.3.2 - Patch 02
VxFS 3.3.2 patch 02 resolves numerous non-critical issues identified in previous versions of the file system (in addition to the five critical issues highlighted below). It has received the most comprehensive broadband testing of any file system version released by VERITAS and provides significant performance improvements. VERITAS therefore encourages all customers to upgrade to version 3.3.2 and then apply patch fs.332-patch02 which will provide better stability, performance and quality. The fs.332-patch02 is a cumulative patch and contains all fixes made since the VxFS 3.3.2 GA release including the fixes from the fs.332-patch01 release (see README.108473-02.txt, README.108474-02.txt, or README.108475-02.txt for more details). Therefore you can upgrade your VxFS 3.3.2 software without first installing fs.332-patch01. However if fs.332-patch01 is already installed on your system you do not need to remove fs.332-patch01 before installing fs.332-patch02.
This patch can also be applied to the following Editions:
* Database Edition 2.1 for Oracle
* Database Replication Option 2.1 for Oracle
* Database Edition 1.3 for Sybase
* File Server Edition 5.0.1
* Web Edition 2.0
These Editions are based on VxFS 3.3.2 and this patch has been verified with the Editions.
Special Note to VxFS 3.3.3 users (or VxFS 3.3.2 customers wanting to upgrade to 3.3.3):
VxFS 3.3.3 is technically the same as VxFS 3.3.2 with patch01 applied, the only difference being that VxFS 3.3.3 also supports Solaris 8. If you are running VxFS 3.3.2 plus patch02, VERITAS recommends that you DO NOT upgrade at this time to VxFS 3.3.3 or you will lose all the fixes provided in fs.332-patch02. VERITAS will release (within a few weeks) patch 01 for VxFS 3.3.3 which will contain the same fixes provided in fs.332-patch02.
Here are the two critical new issues addressed by patch 02:
Issue One (VERITAS Incident Number 39067):
Under certain conditions, a problem could occur on a system that crashed soon after a file system resize or file system reorganization because VxFS was not adequately processing extended operations, resulting in inodes being marked BAD and the following console message:
vxfs: mesg 017: vx_iread_1 - %s file system inode %d marked bad
Note this behavior has been observed only under the rare condition where there was a hardware crash or unclean shutdown during or immediately after a resize or reorg of the file system and has only been reported one time to VERITAS Technical Support.
Issue Two (VERITAS Incident Number 39068):
In certain circumstances. full file system checks were not properly detecting DIRTY file systems. This processing failure lead to corruption on file systems that required extended operations, for example those that were resized.
Note this behavior has been observed only under the rare condition where there was a hardware crash or unclean shutdown during or immediately after a resize or reorg of the file system.
Here are the three critical issues that were addressed in patch 01 and are also incorporated into patch 02:
Issue Three (VERITAS Incident Number 31636):
A problem in the fsck utility (in the function au_state_set()) can corrupt memory. This results in fsck dumping core, in corrupted directory blocks, or in extent map errors reported on the file system after it is mounted. If fsck dumps core before it marks the file system clean, the file system cannot be mounted. This problem can occur only after a system crash (when fsck is run to recover the file system).
File systems with the following characteristics are at risk:
* A Version 3 or Version 4 disk layout
* A file system larger than:
128 GB (1K block size)
256 GB (2K block size)
512 GB (4K block size.
* File systems with an 8K block size are not affected.
* fsck dumps core
* fsck finds corrupted directory blocks or inodes and discards them
* the file system can be mounted, but directory blocks, inode maps, extent maps, or inodes are subsequently found to be "bad," requiring a full file system check.
The following table lets you determine if your file system is at risk:
Fragment Size And "df -g" Reports More Than This Many Blocks
1024 268,435,456 (at risk)
2048 536,870,912 (at risk)
4096 1,073,741,824 (at risk)
8192 (not at risk)
Issue Four (VERITAS Incident Number 31948):
The Inode Allocation Unit (IAU) Summaries are not fixed correctly by fsck process, resulting in the following console message:
vxfs: mesg 004: vx_mapbad - %s file system free inode bitmap in au %d marked bad
and as a further result, requiring a full fsck be performed on the file system. Please note that this problem is rare and only occurs when the IAU structural file is severely fragmented in a particular way, which is likely to occur only on fragmented file systems containing millions of files.
Issue Five (VERITAS Incident Number 33246):
Inodes are marked bad when a system comes back up after a crash or an unclean shutdown. After the system is up and the file systems are all mounted, the following error message will be displayed on the console:
vxfs: mesg 017: vx_iread_1 - %s file system inode 44 marked bad
This problem is due to the inode having extended operations pending during a resize operation. Usually extended operations are completed as part of the mount process, but if a resize operation is performed on the file system when it was last mounted, the extended operation is ignored. Please note we consider this condition rare, and it occurs only when the system crashes or is shutdown incorrectly, following a resize of the file system.
Detailed information on these and other problems resolved by this patch are included in the accompanying README.108473-02, README.108474-02, and README.108475-02 files provided with the patch.
What to do if at Risk:
All previous versions of VERITAS File System need to be upgraded to 3.3.2 and the fs.332-patch02 needs to be applied.
Please Note: If you plan to use VRTSvxfs and Quick I/O (VRTSqio), make sure that both of these packages are installed before installing this patch. And if the patch is installed when only VRTSvxfs is installed and VRTSqio needs to be added later, this patch must first be removed, VRTSqio added, and then the patch re-installed.
Read the file for further information and the location and/or file name of the actual patch.
After downloading, uncompressing, and untarring the patch:
If you are running Solaris 2.5.1:
- cd /3.3.2-patch02/sol2.5.1
- ./108473-02/installpatch 108473-02
If you are running Solaris 2.6:
- make sure you have Sun patch 106125-08 installed
- cd /3.3.2-patch02/sol2.6
- /usr/sbin/patchadd 108474-02
If you are running Solaris 7:
- make sure you have Sun patch 107171-04 installed
- cd /3.3.2-patch02/sol7
- /usr/sbin/patchadd 108475-02
For all Solaris revisions do the following:
-chown root vxdump
-chown root vxrestore
Reboot the system for the patch to take effect.