Change Management Server (CM Server) configuration guidelines for ClearQuest Web

Answer

System Resources

All Platforms
CM Server host must have a minimum of 8 GB of memory.

UNIX

General
At a minimum, the swap space on UNIX host systems must be twice the size of memory.

Solaris SPARC

Solaris 9 -- Add the following to /etc/system:


set shmsys:shminfo_shmmax = 4294967295
set shmsys:shminfo_shmseg = 1024
set shmsys:shminfo_shmmni = 1024
set semsys:seminfo_semaem = 16384
set semsys:seminfo_semmni = 1024
set semsys:seminfo_semmap = 1026
set semsys:seminfo_semmns = 16384
set semsys:seminfo_semmsl = 100
set semsys:seminfo_semopm = 100
set semsys:seminfo_semmnu = 2048
set semsys:seminfo_semume = 256
set msgsys:msginfo_msgmap = 1026
set msgsys:msginfo_msgmax = 65535
set rlim_fd_cur=65535



Solaris 10 -- Run the following commands:


projmod -s -K 'project.max-shm-memory=(privileged,4gb,deny)' user.root
projmod -s -K 'project.max-shm-ids=(privileged,1024,deny)' user.root
projmod -s -K 'project.max-sem-ids=(privileged,1024,deny)' user.root
projmod -s -K 'process.max-sem-nsems=(privileged,512,deny)' user.root
projmod -s -K 'process.max-sem-ops=(privileged,512,deny)' user.root
projmod -s -K 'process.max-file-descriptor=(privileged,65535,deny)' user.root


These settings are persistent in /etc/project.

Solaris 10

• Disable the Nagle algorithm by using the following command:

• Use an AIX Power 5 or Power 6 host system.
• Set the following operating system parameters:

no -o tcp_nagle_limit=0

Most TCP implementations use the Nagle algorithm, where a TCP connection can have only one outstanding small segment that has not been acknowledged. This causes TCP to delay sending further packets until it receives an acknowledgement or until it can bundle more data and send a full-size segment.

You must disable the Nagle algorithm when using the ClearQuest Web application.

no -o clean_partial_conns=1

Specifies whether SYN (synchronize the sequence number) attacks are avoided.

Enable this option for the ClearQuest Web application to protect against network attacks.

no -o tcp_nodelayack=1

Enabling this parameter causes TCP to send immediate acknowledgement (ACK) packets to the sender. When the parameter tcp_nodelayack is off, TCP delays sending ACK packets by up to 200 ms. This allows the ACK packet to piggyback on a response and minimizes the system overhead.

For optimal performance, enable this parameter.

no -o rfc1323=1

Enables window scaling and time stamps as specified by RFC 1323 (TCP Extensions for High Performance). Window scaling allows the TCP window sizes (tcp_recvspace and tcp_sendspace) to be larger than 64 KB (65,536) and is typically used for large maximum transmission unit (MTU) networks.

If your ClearCase VOBs or ClearQuest databases are hosted on a Windows system, enable this parameter for optimal performance.

• Increase the following tunable kernel parameters:


maxfiles=4096
maxfiles_lim=8192
nfile=16384


• Disable the Nagle algorithm by issuing the following command:

ndd -set /dev/tcp tcp_naglim_def 1

• Modify TCP_KEEPALIVE_INTERVAL (Determines the interval between probes)

ndd -set /dev/tcp tcp_keepalive_interval 7200000

default value: none recommended
value: 7200000 (milliseconds)

• Increase JVM VIRTUAL PAGE SIZE (Sets the JVM instruction and data page sizes to 64 MB to improve performance)

For WebSphere 6.1.0.x use the following command

chatr +pi64M +pd64M <WASHOME>/java/bin/PA_RISC2.0/java

default value: 4 MB
recommended value: 64 MB

Stop WebSphere Application Server prior to making this change.

• Modify TCP_CONN_REQUEST_MAX
Specifies the maximum number of connection requests that the operating system can queue when the server does not have available threads. When high connection rates occur, a large backlog of TCP/IP connection requests builds up and client connections are dropped. Adjust this setting when clients start to time out after waiting to connect. Verify this situation by issuing the netstat -p tcp command. Look for "connect requests dropped due to full queue."

ndd -set /dev/tcp tcp_conn_request_max 8192

default value: 4096
recommended value: 8192

• serverWorkerThreadCount - Default value is 30.

Description : The number of threads that a ClearQuest RPC process will use to handle incoming requests.

Configuration : Increasing this value allows for a fewer number of ClearQuest RPC processes. Update this value with caution because setting it to an improper value can have a negative effect on performance. Setting this value too high will cause contention within a process, and setting this value too low increases the need for more ClearQuest RPC processes.

• activeHttpSessionThreshold - Default value is 30.

Description : The number of active HTTP sessions on a ClearQuest RPC process. If this value is exceeded, it may cause the CM Server to start another ClearQuest RPC process.

Configuration: Increasing this value will allow for fewer ClearQuest RPC processes. Update this value with caution, as setting it too high will cause contention within a process, and setting it too low will increase the need for more ClearQuest RPC processes.

• maximumActiveServers - Default value is 10.

Description : The maximum number of active ClearQuest RPC processes allowed by the CM Server. This number may be adjusted to suit the host system’s resource capacity.

• recycleServerHttpSessionLimit - Default value is 200. A value of zero indicates that this limit will not be checked.

Description : The number of HTTP sessions that a ClearQuest RPC process will service before it is recycled by the CM Server.

• recycleServerLifetimeLimit - Default value is 86,400 seconds (24 hours). A value of zero indicates that this limit will not be checked.

Description : The maximum life time, in seconds, of a ClearQuest RPC process before it is recycled by the CM Server.

• recyclingPeriod - Default value is 3,600 seconds (60 minutes). This value must be greater than zero.

Description : The amount of time, in seconds, that a recycling ClearQuest RPC process is allowed to shutdown and complete any outstanding session work before it terminates.

Configuration: Update this value with caution, as setting this to an improper value can have a serious effect on performance. This value overrides the behaviors of the following TeamServerMBean attributes: sessionObjectDefaultTimeout, sessionQueryObjectTimeout, and sessionRecordObjectTimeout.


• remoteChartReportServerHostName - Default value is "" (empty string).
  
  Description : Applicable on UNIX CM Server systems only. Designates the host name of the Windows CM Server used for processing chart and report requests. Added in ClearQuest version 7.1.1.
  
• remoteChartReportServerPort - Default value is 0.
  
  Description : Applicable for Unix CM Server systems only. Designates the port number of the Windows CM Server used for processing chart and report requests. Added in ClearQuest version 7.1.1.
  

• J2C Connection pool - Default value is 400.
  
  Description : The maximum number of connections to cqrpc backend processes.
  
  Configuration : Increase the J2C Connection pool value "Maximum Connections" if the average number of CQWeb users is greater than the current value of "Maximum Connections".
  
  The Maximum Connections value is located at: Applications > Enterprise Applications > TeamEAR > Manage Modules > OncrpcConnector > Resource Adapter > J2C connection factories > CqRpc > Connection pool properties