Solaris Cluster Global Filesystem is Slow/Hang when Writing a lot of Small Files

(Doc ID 1321377.1)

Last updated on NOVEMBER 07, 2017

Applies to:

Solaris Cluster - Version 3.2 to 3.3 U1 [Release 3.2 to 3.3]
Oracle Solaris on SPARC (64-bit)
Oracle Solaris on x86-64 (64-bit)

Symptoms

If using Solaris Cluster with global filesystem and writing a lot of small files to it the response of the global filesystem can be slow.

The following script writes the result of date command on Solaris Cluster global filesystem (pxfs).
Script:

#!/bin/sh

LOGDIR=`pwd`                            

LOGFILE=${LOGDIR}/write_test.log

presec=`date '+%S'`
while sun;do
sec=`date '+%S'`

delta=`expr ${sec} - ${presec}`
test ${delta} -lt 0 && delta=`expr ${delta} + 60`
warn=""
test ${delta} -gt 1 && warn="WARN"

printf "`date` ${delta} ${warn}\n" >> ${LOGFILE}
presec=${sec}
sleep 1
done

The script generates a large number of small writes.
40 directories were created on each node on the same pxfs with script copied to each directory.  All script was ran at the same time.

 

Changes

 

Cause

Sign In with your My Oracle Support account

Don't have a My Oracle Support account? Click to get started

My Oracle Support provides customers with access to over a
Million Knowledge Articles and hundreds of Community platforms