IT24124: FOR MQ FOR NONSTOP ON ITANIUM 5.3.1, EC REPEATEDLY RESTARTS CERTAIN CHILD PROCESS (LQMA).
Closed as program error.
EC repeatedly restarts certain child process (LQMA) which lead to high CPU usage. The EC gets into a loop and spawns new LQMAs trying a process name that is already in use. This is a situation the EC is not prepared to handle as keeping a list of free numbers and this number is part of the free list. EC is unable to find a way out of this loop as long as the process name stays is in use. This process spawn loop produces a lot of load on the system. This APAR is working on an EC enhancement to bypass a process name if a process start fails because of a name conflict.
The customer should make sure there are no processes matching the agent process name rule. If it happens, the customer could check if the process using the conflicting process name can be stopped and restarted with a different process name.
Root cause is a wrong error handling inside the EC in the case of a failed process start. The EC receives EAGAIN and _TPC_BADNAME_FERROR on process completion and tries the same name again.
The error handling in the case of an agent process creation failure has been modified. A process name that just caused a bad name error will not be used on the next attempt.
Don't use process names for non-MQ processes that matches the naming rule defined by the user in qmproc.ini.
Reported component name
WEBS MQ NSS ITA
Reported component ID
NoSpecatt / Xsystem
Last modified date
APAR is sysrouted FROM one or more of the following:
APAR is sysrouted TO one or more of the following:
Fixed component name
WEBS MQ NSS ITA
Fixed component ID
Applicable component levels
Translate this page: