|
By the time you find out that a connection is in
a FAILING or DISCONNECTING state, it is likely that it had been in
that state for a while. However, if the connection recently went
into a FAILING or DISCONNECTING state (for example, you just recycled
a subsystem), then you should ensure that enough time has elapsed
to allow the FAIL and DISCONNECT events to be processed by the active
connectors. Normally, this takes no more than a few minutes. IXL040E
may be issued to identify a response that a connector has not provided.
- Determine if the FAILING or DISCONNECTING connection event is
owed a response.
Issue the following command for each ACTIVE connection: DISPLAY
XCF,STRUCTURE,STRNAME=strname,CONNAME=conname
Message
IXC360I will provide detailed information for the connector specified
in the command. If the connector is owed a response (such as a response
to a FAIL or DISCONNECT event), the detailed information might include
the following:
CONNECTION ID(s) OWING A RESPONSE FOR THIS CONNECTOR:
id id id... id
The IDs owing a response to the FAILING
or DISCONNECTING connection event identify the connector (job) that
is not responding. Those jobs owing a response are candidates for
recovery actions. If CONNECTION ID(s) OWING A RESPONSE was
included in the IXC360I output, check the health of the job or system
to see if there is anything obviously wrong with the job or system
on which the job resides. If so, go to Determine recovery actions. Otherwise,
continue to find information about the structure itself.
- Display all relevant information about the structure that might
identify connectors or systems requiring recovery actions.
DISPLAY
XCF,STRUCTURE,STRNAME=strnameCONNAME=ALL
For
each ACTIVE connector (as displayed in IXC360I), check the health
of the connector (job), and check the health of the system on which
the ACTIVE connector resides.
- Issue commands that are processed by the ACTIVE connector. For
example, if the connector is an IRLM job, issue: F IRLMx,STATUS
F IRLMx,STATUS,ALLD
If the job owing the ACTIVE connector
does not respond as normal, then it is a candidate for recovery actions.
- Issue commands on the system on which the ACTIVE connector resides.
For example: D T
If the system does not respond
as normal, then it is a candidate for recovery actions.
- Issue DISPLAY XCF,SYSPLEX,ALL to ensure that all systems
are ACTIVE in the parallel sysplex.
- Issue DISPLAY R,L,CN=(ALL) to see if there are any
outstanding events that might be holding up the processing of the
FAILING or DISCONNECTING connection.
- Examine syslogs for signs of trouble such as ABENDs or DUMPs around
the suspected time of the connector hang.
If a job owning an ACTIVE
connector has experienced ABENDs or DUMPs, then it is a candidate
for recovery actions. If the system has experienced problems such
as ABENDs, DUMPs, SPINLOOPs, or other anomalies, then it is a candidate
for recovery actions.
|