If you suspect that a PDSE is failing, issue the ANALYSIS command
from each system where the PDSE has been reported as failing. Do not
issue the FREELATCH command without preceding it with an ANALYSIS
command.
These variables indicate the kinds of message information:
- nnnnnnnn
- Number of seconds a lock or latch is held.
- aaaa
- ASID (address space identifier) of the latch holder.
- tttttttt
- TCB address of the latch holder. If an SRB holds the latch, the
address is greater than 01000000.
- llllllll
- Address of the latch that is held.
- ssssssss
- System to which a message is directed.
- hhhhhhhh
- Address of the HL1B that is the control block that contains the
locking information for a PDSE.
- dddddddd
- Address of the DIB that contains the data structure information
for a PDSE.
- asrbasrb
- Address of the ASRB that represents the users address space in
the PDSE address space.
- licblicb
- Address of the LICB which is the root control block for PDSE locking
operations.
- fibfibfi
- Address of the FIB which is the root control block for a member
of a PDSE.
- jjjjjjjj
- Job name for the apparent hold of a latch or lock.
- stststst
- Started task name for the apparent holder of a latch or lock.
The SMXC is the locking address space for the PDSE. You should never
force or cancel the SMXC.
- nummsgs
- Number of outstanding messages from this system.
- n holders
- Number of holders for a lock.
- n writers
- Number of writers for a lock.
You should see one of the following results:
- ++ No PDSEs connected
- ++ No exceptional data set conditions detected
- This indicates that the exception detected is either temporary
in nature or is not one of the class of problems that this tool is
designed to detect or fix. If message IGW038A continues to occur,
obtain an SVC dump of the SMSPDSE(1) address space and contact your IBM® service representative.
- ++ Message to ssssssss pending for nnnnnnnn seconds
- This system has been waiting for another system to respond to
it. Try running the ANALYSIS command on the other system. The problem
is probably on the other system. This command is sometimes displayed
with the DIB latch message. When the other system fails to respond
during an OPEN, both messages will appear, but the problem is still
on the other system and no attempt should be made to issue the FREELATCH
command to free the latch on this system.
- ++ Unable to latch DIB:dddddddd
- LATCH:llllllll Holder (aaaa:
tttttttt )
- Holding JOB:jjjjjjjj | Holding Started
Task (stststst |
- Unable to determine job or started task name for holder
- This latch is held only during OPEN and should not be held for
very long, unless the system is waiting for a response from another
system. The latch blocks new OPENs and CLOSEs for the data set in
question. You should first attempt to cancel and force the job or
user that has the latch. If that does not work, issue a FREELATCH
command for this data set. If that does not work, either re-IPL
or restart the PDSE address space.
- ++ Unable to latch HL1B:hhhhhhhh
- LATCH: Holder ( aaaa: tttttttt )
- Holding JOB:jjjjjjjj | Holding Started
Task (stststst |
- Unable to determine job or started task name for holder
- This latch is held only during an attempt to serialize updates
to the PDSE. The latch should never be held for very long because
it blocks most updates and reads of the data set in question. You
should first attempt to cancel and force the job or user that has
the latch. If that does not work, issue a FREELATCH command for this
data set. If that does not work, either re-IPL or restart the
PDSE address space.
- ++ Unable to latch HL1BPLCH:hhhhhhhh
- LATCH:llllllll Holder (aaaa:
tttttttt )
- Holding JOB:jjjjjjjj | Holding Started
Task (stststst |
- Unable to determine job or started task name for holder
- This latch is held only during an attempt to send messages to
another system. The latch should never be held for very long, unless
a message is being sent to another system. The latch blocks most updates
and reads of the data set in question. If any messages to another
system are outstanding, you should perform analysis and repair on
that system. If no messages to other systems are outstanding, you
should first attempt to cancel and force the job or user that has
the latch. If that does not work, issue a FREELATCH command for this
data set. If that does not work, either re-IPL or restart the
PDSE address space.
- ++ Unable to latch ASRBULCH:asrbasrb
- LATCH:llllllll Holder (aaaa: tttttttt )
- Holding JOB:jjjjjjjj | Holding Started
Task (stststst) |
- Unable to determine job or started task name for holder
- This latch is only held while the address space and task level
lock recovery chains are being updated. This will prevent other PDSE
operations in the same address space from proceeding. You should
first attempt to cancel and force the job or user that has the latch.
If that does not work, issue a FREELATCH command for this latch. If
that does not work, either re-IPL or restart the PDSE address space.
- ++ Unable to latch LICBASRBCHAINLATCH:licblicb
- LATCH:llllllll Holder (aaaa: tttttttt )
- Holding JOB:jjjjjjjj | Holding Started
Task (stststst) |
- Unable to determine job or started task name for holder
- This latch is only held on the first PDSE request for a PDSE in an address space or at EOM for
the same address space. It is used to serialize the acquiring and freeing of address space PDSE
control blocks. You should first attempt to cancel and force the job or user that has the latch. If
that does not work, issue a FREELATCH command for this latch. If that does not work, either re-IPL
or restart the PDSE address space.
- ++ Unable to latch ASRBASCB HASH TABLE LATCH:licblicb
- LATCH:llllllll Holder (aaaa:
tttttttt )
- Holding JOB:jjjjjjjj | Holding Started Task
(stststst) |
- Unable to determine job or started task name for holder
- This latch is obtained when an address space first utilizes PDSE resoruces, when an address
space shuts down and when the PDSE1 address space restarts to determine which address spaces may
have to reconnect to PDSE data sets and members.
- ++ Unable to latch FIBHashTableLATCH:dddddddd
- LATCH:llllllll Holder (aaaa: tttttttt )
- Holding JOB:jjjjjjjj | Holding Started
Task (stststst) |
- Unable to determine job or started task name for holder
- This latch serializes the access to members of a PDSE. It can
prevent most access of a PDSE from being performed. If will only be
identified when a analysis command is directed to a single PDSE.
You should first attempt to cancel and force the job or user that
has the latch. If that does not work, issue a FREELATCH command for
this latch. If that does not work, either re-IPL or restart the PDSE
address space.
- ++ Unable to latch FIB_LATCH:fibfibfi
- LATCH:llllllll Holder (aaaa: tttttttt )
- Holding JOB:jjjjjjjj | Holding Started
Task (stststst) |
- Unable to determine job or started task name for holder
- This latch serializes the access to a single member of a PDSE.
If will only be identified when a analysis command is directed to
a single PDSE. You should first attempt to cancel and force the job
or user that has the latch. If that does not work, issue a FREELATCH
command for this latch. If that does not work, either re-IPL or restart
the PDSE address space.
- ++ Unable to latch DIB:llllllll Holders
changing
- ++ Unable to latch HL1B:llllllll Holders
changing
- ++ Unable to latch HL1BPLCH:llllllll Holders
changing
- This is an informational message. It indicates that although the
latch was held every time you tried to access the structures, the
holder was dynamic. This indicates high activity on the data set but
does not indicate any error. The occurrence of any of these messages
is unlikely. They are here only for information.
- ++ Unable to obtain latch for DIB | HL1B HashTable:llllllll Holder(aaaa:tttttttt )
- This message indicates that a global latch that is used to locate
the structures for the PDSE is frozen. You should attempt to cancel
the job or user that is associated with the ASID. If the system still
holds the latch, force the initiator. If the system still holds the
latch, you should attempt the FREELATCH command for this latch.
- ++ Unable to obtain latch for DIB | HL1B HashTable:llllllll Holders
changing
- This message indicates that a global latch used to locate the
structures for the PDSE is very busy, preventing successful completion
of the analysis. Reissue the FREELATCH command until the analysis
is successful.
- ++ Unable to latch XCM Holder(aaaa:tttttttt )
- The SMXC address space obtains this latch only when XCF messages
are sent to another system or when messages are received from another
system. This indicates that a global latch that is not associated
with a particular data set is frozen. You should attempt to issue
the FREELATCH command for this latch.
- ++ Unable to latch XCM Holders changing
- This message indicates that a global latch used to serialize messaging
between systems is very busy, preventing successful completion of
some analysis. You should reissue the FREELATCH command until the
analysis is successful.
- ++ Latch DIB | HL1B | HL1BPLCH Nullified
- ++ Latch DIB | HL1B | HL1BPLCH Broken
- This message indicates that the specified latch either is broken
or has been nullified. Both cases indicate that the data structures
for the PDSE are compromised. You should attempt to have all users
unallocate the PDSE; this might clean up the data structures.
- ++ Lock GLOBAL | LOCAL DIRECTORY | FORMATWRITE SHARED | EXCLUSIVE
- Held for at least nnnnnnnn seconds
- HL1B:hhhhhhhh HOLDER(aaaa:tttttttt)
- Holding JOB:jjjjjjjj | Holding Started
Task:stststst |
- Unable to determine job or started task name for holder
-
This message indicates that the system has held a directory
or format write lock for a long time. The indicated amount of time
begins with the first ANALYSIS command that detected the waiting lock
request.
If the lock is GLOBAL, it has been granted for the
sysplex. If there is contention on this resource, you might want to
cancel the job that holds the lock.
If the lock is held locally,
this system is waiting for another system to release the lock, so
that it can be granted. If there is no message to another system,
you might need to cancel this job. If there is a message to another
system, issue the ANALYSIS command on that system and resolve the
problem on that system.
If the command cannot detect the lock
holder, you might be able to fix the problem by canceling all users
of the data set.
- ++ n holders Additional holders of DIRECTORY
|FORMAT WRITE Lock
- ++ n writers waiting for DIRECTORY |
FORMAT WRITE LOCK Shared | Exclusive
- These messages give an indication of the amount of activity against
a data set when a lock has been held for an excessive amount of time.
- ++ nummsgs unresponded messages to ssssssss
- This is a summary of messages sent to other systems that have
been outstanding for too long. If you do not find any other errors,
run the ANALYSIS command on the other system.
- ++ Broken PDSE structure —
- Bad HL3B | LRE encountered
- Internal structures of the PDSE are broken. Attempt to quiesce
the use of the PDSE. This frees all existing PDSE structures and might
resolve the PDSE problem.