Solaris 10 systems may experience performance degradation when deleting large amounts of data. During this time, the output of "df -k" may display incorrect numbers in the "used" and "capacity" columns.

This issue can occur in the following releases:

SPARC Platform

• Solaris 10 without patch 118833-18

x86 Platform

• Solaris 10 without patch 118855-15

Note 1: Solaris 8 and Solaris 9 are not impacted by this issue.

Note 2: In most cases, the removal of three 5G files will trigger this issue.

The described issue only occurs if UFS logging is enabled. To determine if UFS logging is enabled, the following command can be used:

    % mount | grep logging
    ...
    /data on /dev/md/dsk/d10 read/write/setuid/devices/intr/largefiles/logging/xattr/onerror=panic/dev=154000a 
    on Wed Mar 22 16:40:21 2006

Note: UFS logging is enabled by default.

During large file deletion on a logging UFS, performance will be impacted due to an increase in kernel cpu consumption as seen in the "sy" column of vmstat(1M). The output from df(1M) during this time will have large numbers in the "used" and "capacity" columns and then return to normal.

On affected systems, the output from the df(1M) command will be similar to the following:

    % /bin/df -k
    ...
    /dev/md/dsk/d10    8707103 18446744073708818053 9353595 213998556776922% /data

There is no workaround. Please see the "Resolution" section below.

This issue is addressed in the following releases:

Solaris Platform

• Solaris 10 with patch 118833-18 or later

x86 Platform

• Solaris 10 with patch 118855-15 or later

• Updated Contributing Factors section

• State Resolved
• Updated Contributing Factors and Resolution sections

This solution has no attachment