Sun Microsystems, Inc.  Oracle System Handbook - ISO 7.0 May 2018 Internal/Partner Edition
   Home | Current Systems | Former STK Products | EOL Systems | Components | General Info | Search | Feedback

Asset ID: 1-79-1903388.1
Update Date:2017-12-20
Keywords:

Solution Type  Predictive Self-Healing Sure

Solution  1903388.1 :   SuperCluster - ssctuner reference document  


Related Items
  • Sun ZFS Storage 7320
  •  
  • Oracle SuperCluster M6-32 Hardware
  •  
  • Oracle SuperCluster M7 Hardware
  •  
  • Oracle Exadata Storage Server Software
  •  
  • Oracle SuperCluster M7 Hardware
  •  
  • Oracle SuperCluster Specific Software
  •  
  • Oracle SuperCluster T5-8 Half Rack
  •  
  • SPARC SuperCluster T4-4 Half Rack
  •  
  • Solaris Operating System
  •  
  • Solaris Cluster
  •  
  • Oracle SuperCluster T5-8 Full Rack
  •  
  • SPARC SuperCluster T4-4 Full Rack
  •  
Related Categories
  • PLA-Support>Eng Systems>Exadata/ODA/SSC>SPARC SuperCluster>DB: SuperCluster_EST
  •  
  • _Old GCS Categories>ST>Server>Engineered Systems>SPARC SuperCluster>Install
  •  
  • _Old GCS Categories>ST>Server>Engineered Systems>SPARC SuperCluster>Patching
  •  


SuperCluster tuning assistant

In this Document
Purpose
Scope
Details
 Definition
 How to install it
 How to run it "live" making recommended tuning changes
 How to change or disable components of what ssctuner does/checks
  To make critical messages go to your email address, for example
 To change disk usage ( / and zone roots ) warning level to 80% 
 To enable thread priority changing for non-exa Oracle DB domains
 To enable zpool check/repair of vdisk zpools not generated by the SuperCluster installer
 To exclude NFS mounts from warning mechanism
 To include NFS mounts in warning mechanism (overrides exclude )  
 To change the amount of time ssctuner will suppress repeat warnings for 
 Updating
 
 Legacy Information
 Generic feedback
References


Applies to:

Oracle SuperCluster M6-32 Hardware - Version All Versions to All Versions [Release All Releases]
Oracle SuperCluster T5-8 Half Rack - Version All Versions to All Versions [Release All Releases]
SPARC SuperCluster T4-4 Half Rack - Version All Versions to All Versions [Release All Releases]
Oracle Exadata Storage Server Software - Version 11.2.3.2.1 to 11.2.3.2.1 [Release 11.2]
SPARC SuperCluster T4-4 Full Rack - Version All Versions to All Versions [Release All Releases]
Oracle Solaris on SPARC (64-bit)
Linux x86-64

Purpose

Describe the purpose and functionality of the mandatory ssctuner service on Oracle SuperCluster systems

Scope

 Will cover the version of ssctuner starting with the JAN 2014 Quarterly Full Stack Download Patch (QFSDP)

Details

Definition

ssctuner is a small set of Perl/Korn shell scripts and config files
designed to be run on Oracle SuperCluster Solaris 10 and 11 global
zones only.  It runs as an SMF service to tune /etc/system, ndd
parameters, /kernel/drv/sd.conf, /kernel/drv/ssd.conf, and
periodically check for other things like use of DISM or suboptimal NFS
mount options.  These items are checked every 2 hours by default.
Note that, if you manually tune a parameter ssctuner requires to have
a different value, it will set the value of that parameter back to
what it requires and log the fact that it made changes at this
interval check.

ssctuner checks to make sure the package (IPS or SvR4) is not
more than 1 year out of date and issues a warning once it is more
than 8 months since the last update, and a critical message once
it is more than a year old.  Though ssctuner is normally delivered
only as part of an install or Quarterly Full Download Software Patch,
this check is specific to the version of ssctuner.


It also checks, every 2 minutes, to see if there are any virtual disk
devices in the degraded state that have come back online and will clear
that zpool (and attempt to repair that fmadm fault in 11.1) if so.

Solaris 11 must be at SRU 11.4 or later for this tool or  ssd.conf/sd.conf settings could cause panics. Also, if you currently
set ndd parameters via another SMF service or init script, that must be removed.  ssctuner should manage the ndd parameters.

How to install it

ssctuner should be automatically installed as part of SuperCluster
infrastructure configuration. scctuner follows standard s11 and s10 package 
install practices and no special install instructions/steps are required 
anymore.  The Solaris 11 package is ssctuner and resides in the exa-family
repository.  The Solaris 10 package is ORCLssctuner.

How to run it "live" making recommended tuning changes

It is enabled as part of install, if it is ever disabled, enable again by:

#svcadm enable ssctuner

 

Note that /etc/system, /kernel/drv/sd.conf and /kernel/drv/ssd.conf  parameters will not take effect until a reboot occurs after enabling
the service... So check that current date minus uptime is earlier than timestamp in /etc/system comment to know that they are in effect (or
just reboot if convenient to be sure).

How to change or disable components of what ssctuner does/checks

With 1.1.4 changes, ssctuner now uses SMF property variables to
control features.  Use standard svcprop (note this will show only
default settings not changes) and svccfg tools to view and set
ssctuner_vars variables.

Changing certain ssctuner SMF properties like EMAIL_ADDRESS and
disk or memory usage warning levels may help for some deploy-
ments. Disabling individual ssctuner service components can
have unpredictable consequences; no changes should be made to
these properties except under instruction from oracle support.
The ssctuner service should never be disabled or uninstalled.

NOTE : EVEN THOUGH WE SHOW YOU EXAMPLES PLEASE CHANGE THESE ONLY IN CONSULTATION WITH ORACLE SUPPORT!

  

# svccfg -s ssctuner listprop 'ssctuner_vars/*'
ssctuner_vars/CRIT_THREADS_FIX          boolean     true
ssctuner_vars/CRIT_THREADS_NONEXA       boolean     false
ssctuner_vars/DISK_SPACE_CHECK          boolean     true
ssctuner_vars/DISK_USAGE_CRIT           integer     90
ssctuner_vars/DISK_USAGE_WARN           integer     85
ssctuner_vars/DISM_CHECK                boolean     true
ssctuner_vars/EMAIL_ADDRESS             astring     root@localhost
ssctuner_vars/EMAIL_MESSAGES            boolean     true
ssctuner_vars/FORCELOAD_VDC             boolean     false
ssctuner_vars/INTRD_DISABLE             boolean     true
ssctuner_vars/ISCSI_TUNE                boolean     true
ssctuner_vars/MAJOR_INTERVAL            integer     120
ssctuner_vars/MEM_USAGE_CRIT            integer     97
ssctuner_vars/MEM_USAGE_WARN            integer     94
ssctuner_vars/MINOR_INTERVAL            integer     2
ssctuner_vars/NDD_TUNE                  boolean     true
ssctuner_vars/NFS_CHECK                 boolean     true
ssctuner_vars/NFS_EXCLUDE               astring
ssctuner_vars/NFS_INCLUDE               astring
ssctuner_vars/NTPCONF_TUNE              boolean     true
ssctuner_vars/POWERADM_DISABLE          boolean     true
ssctuner_vars/SDCONF_TUNE               boolean     true
ssctuner_vars/SERD_THRESHOLD_TUNE       boolean     true
ssctuner_vars/SSDCONF_TUNE              boolean     true
ssctuner_vars/SYSLOG_DUP_SUPPRESS_HOURS integer     8
ssctuner_vars/SYSTEM_TUNE               boolean     true
ssctuner_vars/ZPOOL_FIX                 boolean     true
ssctuner_vars/ZPOOL_NAME_CUST           astring 

 To make critical messages go to your email address, for example

# svccfg -s ssctuner setprop ssctuner_vars/EMAIL_ADDRESS="my_name@mycorp.com" 

To change disk usage ( / and zone roots ) warning level to 80% 


# svccfg -s ssctuner setprop ssctuner_vars/DISK_USAGE_WARN=80 

To enable thread priority changing for non-exa Oracle DB domains

# svccfg -s ssctuner setprop ssctuner_vars/CRIT_THREADS_NONEXA=true 

To enable zpool check/repair of vdisk zpools not generated by the SuperCluster installer


# svccfg -s ssctuner setprop ssctuner_vars/ZPOOL_NAME_CUST=my_vdisk_pool 

To exclude NFS mounts from warning mechanism

# svccfg -s ssctuner setprop ssctuner_vars/NFS_EXCLUDE='mount_name_or_device' 

To include NFS mounts in warning mechanism (overrides exclude )  

# svccfg -s ssctuner setprop ssctuner_vars/NFS_INCLUDE='mount_name_or_device'
 
NOTE : NFS_EXCLUDE, NFS_INCLUDE and ZPOOL_NAME_CUST should be kept simple strings but can be simple regular expressions.  If you need the flexibility of
regular expressions, be extremely careful to double quote and verify ssctuner service comes back after restarting with no errors in SMF log file or you
may end up preventing the ssctuner service from functioning.  One example of what is not legal :
svccfg -s ssctuner setprop ssctuner_vars/NFS_EXCLUDE=zurich\|jurassic
svccfg -s ssctuner setprop ssctuner_vars/NFS_EXCLUDE='zurich|jurassic'
  
And what is legal:
svccfg -s ssctuner setprop ssctuner_vars/NFS_EXCLUDE='zurich\|jurassic'
  

To change the amount of time ssctuner will suppress repeat warnings for 


# svccfg -s ssctuner setprop ssctuner_vars/SYSLOG_DUP_SUPPRESS_HOURS=24

 

If you use the wrong syntax svcs  ssctuner will show the service not come back
and the following line will show the offending property you must fix :
#grep -i parameter /var/svc/log/site-application-sysadmin-ssctuner:default.log
  

After any changes are made, the service must be refreshed and restarted to take effect :
# svcadm refresh ssctuner
# svcadm restart ssctuner
  

 

EXCEPT UNDER INSTRUCTION FROM ORACLE SUPPORT. The ssctuner service should never be disabled or uninstalled.

If ever directed to the way to disable ssctuner is 

#svcadm disable ssctuner

  

This will stop the repeated checks but leave most  tuning changes in tact (a reboot after this will lose ndd tunings)

Updating

ssctuner is updated along with the exafamily updates with each QFSDP.

It is permitted to update ssctuner to a version newer than the existing QFSDP if directed to by support.

However you should not run and older version compared to your QFSDP level. 

When ssctuner package is updated on Solaris 11, new features may be added
along with new names  for portions that can be disabled if needed. 

Those will always be visible in the SMF properties, and changes to SMF property
changes in Solaris 11 will persist across ssctuner package upgrades, though
not if the ssctuner pkg is uninstalled and reinstalled.


Legacy Information

The legacy /opt/oracle.supercluster/ssctuner/etc/ssctuner.conf will be
read after ssctuner SMF properties and can override them but this
mechanism is deprecated and will go away in a future release.  It can
have settings for each component, allowing one to disable portions of
what ssctuner does while keeping other portions active.  Note that
disabling individual ssctuner service components can have
unpredictable consequences; no changes should be made to this
configuration file

CRIT_THREADS_FIX=TRUE
DISM_CHECK=TRUE
FORCELOAD_VDC=FALSE
INTRD_DISABLE=TRUE
ISCSI_TUNE=TRUE
NDD_TUNE=TRUE
NFS_CHECK=TRUE
POWERADM_DISABLE=TRUE
SDCONF_TUNE=TRUE
SSDCONF_TUNE=TRUE
SYSTEM_TUNE=TRUE
ZPOOL_FIX=TRUE



 

How to turn off and remove the service undoing all tuning changes

To remove ssctuner from the system, use standard Solaris package removal commands. A script "rollback.ksh" is available to undo any system configuration changes.

This script needs to be run before the package removal and will rollback any changes to the status prior to the current ssctuner install (including system changes not related to ssctuner).

The script runs in interactive mode allowing the user to select which changes to rollback. Note that /etc/system parameters will persist in effect until a reboot occurs after uninstalling.

  


 

Generic feedback

For general feedback about the Oracle SuperCluster ssctuner service please fell free to discuss in the MOS communities Oracle SuperCluster General Tuning Discussion.

References

<NOTE:1567979.1> - Oracle SuperCluster Supported Software Versions - All Hardware Types
<NOTE:2008049.1> - Oracle SuperCluster SSCtuner Best Practices

Attachments
This solution has no attachment
  Copyright © 2018 Oracle, Inc.  All rights reserved.
 Feedback