gpssh-exkeys ganglia perf stat

python /usr/local/greenplum-db/./bin/gpssh-exkeys -f host 
perf stat 
http://blog.chinaunix.net/uid-11121450-id-3147002.html
开源监控软件ganglia安装手册

[root@mdw ~]# more host 
mdw
sdw1
sdw2
sdw3
sdw4
[root@mdw ~]# gpssh-exkeys  -f host
[STEP 1 of 5] create local ID and authorize on local host
  ... /root/.ssh/id_rsa file exists ... key generation skipped


[STEP 2 of 5] keyscan all hosts and update known_hosts file


[STEP 3 of 5] authorize current user on remote hosts
  ... send to sdw1
  ... send to sdw2
  ... send to sdw3
  ... send to sdw4


[STEP 4 of 5] determine common authentication file content


[STEP 5 of 5] copy authentication files to all remote hosts
  ... finished key exchange with sdw1
  ... finished key exchange with sdw2
  ... finished key exchange with sdw3
  ... finished key exchange with sdw4


[INFO] completed successfully
[root@mdw ~]# 
[root@mdw ~]# gpssh -f host
=> pwd
[sdw4] /sys/kernel/mm/redhat_transparent_hugepage
[sdw3] /sys/kernel/mm/redhat_transparent_hugepage
[ mdw] /sys/kernel/mm/redhat_transparent_hugepage
[sdw2] /sys/kernel/mm/redhat_transparent_hugepage
[sdw1] /sys/kernel/mm/redhat_transparent_hugepage
=> cat ena*
[sdw4] [always] never
[sdw3] [always] never
[ mdw] [always] never
[sdw2] [always] never
[sdw1] always [never]
=> echo never >enabled*
[sdw4]
[sdw3]
[ mdw]
[sdw2]
[sdw1]
=> cat ena*
[sdw4] always [never]
[sdw3] always [never]
[ mdw] always [never]
[sdw2] always [never]
[sdw1] always [never]
=> exit
[root@mdw ~]# gpssh -f host
Note: command history unsupported on this machine ...
=> echo "echo never >/sys/kernel/mm/redhat_transparent_hugepage/enabled"  >/etc/rc.local
[sdw4]
[sdw3]
[ mdw]
[sdw2]
[sdw1]
=> cat /etc/rc.local
[sdw4] echo never >/sys/kernel/mm/redhat_transparent_hugepage/enabled
[sdw3] echo never >/sys/kernel/mm/redhat_transparent_hugepage/enabled
[ mdw] echo never >/sys/kernel/mm/redhat_transparent_hugepage/enabled
[sdw2] echo never >/sys/kernel/mm/redhat_transparent_hugepage/enabled
[sdw1] echo never >/sys/kernel/mm/redhat_transparent_hugepage/enabled
=> exit
echo never > /sys/kernel/mm/redhat_transparent_hugepage/enabled
echo never > /sys/kernel/mm/redhat_transparent_hugepage/defrag


RHEL 6.2 and 6.3 include a feature called "transparent hugepage compaction" which interacts poorly with Hadoop workloads. This can cause a serious performance regression compared to other operating system versions on the same hardware.


Symptom: top and other system monitoring tools show a large percentage of the CPU usage classified as "system CPU". If system CPU usage is 30% or more of the total CPU usage, your system may be experiencing this issue.


Bug: https://bugzilla.redhat.com/show_bug.cgi?id=805593
Severity: Medium (up to 3x performance loss)
Anticipated Resolution: Currently working with Red Hat to resolve for a future RHEL update
Workaround: Add the following command to /etc/rc.local to disable transparent hugepage compaction:


解决方法：
echo always > /sys/kernel/mm/redhat_transparent_hugepage/enabled
echo never > /sys/kernel/mm/redhat_transparent_hugepage/defrag