A fix is available
APAR status
Closed as program error.
Error description
Your application is designed such that it performs dynamic routing to target regions. Once the tasks in the target regions begin running, they may also perform a dynamic DPL. In this scenario, the running task may now have two CPSM WLNE control blocks allocated to it - one for the initial arrival of the routed work (route initiate, RTINIT) and a second for the dynamic DPL request (route select, RTSELECT). . Under this environment, if MXT is set high enough, all available WLNEs in the pool become used. Because of the lack of free WLNEs in the pool, CPSM performs garbage collection to free up orphaned elements, if any exist. . When garbage collection finds that no free elements exist, it issues the following message and produces a diagnostic dump: . EYUWG0106E WLM has encountered an error while attempting to release MAS resources. . The stack for the task that takes the dump will look similar to the following . Task Meth Load-Pt OSSB Stack Mal MODB MOEB 12345 XLOP 000AEE28 3F91ABB0 3F91ABD8 00000000 29B20000 28D7C140 12345 WNGC 29D4A380 3F91ABB0 3F91AD20 2B3A5E0C 29B2BB00 28D7CA10 12345 WNLM 29D3F650 3F91ABB0 3F91AFE0 3F91AE30 29B2BB00 28D7CA10 12345 XZSD 2716D6F8 29B2EE00 29B2EF70 3F91B120 29B21100 28D7C460 . To diagnose that this issue is the cause of the message and dump, and that it is NOT a case of a long running task obtaining WLNEs but not freeing them (RTINIT calls with no RTTERMINATE calls) you need to examine the individual WLNEs. . The WLNEs are located in the MAS regions private storage. Starting with the WLM MOEB, MOEB_MAS_LISTHD points to the list header. In the list header, WNLH_WNLE points to the first element. The elements are contiguous in storage, and each begins with the owning taskid. Visually examine each of the elements to verify that tasks are taking two elements, and that one task is not using the majority. The number of elements to examine is in field WNLH_CELL_MAX in the header. Currently that is 1024 or x'400'. . Note that even though the EYUWG0106E message is issued, and a diagnostic dump is taken, routing will still continue as normal. CPSM will dynamically allocate then free WLNE elements that are outside of the pool. This may add additional overhead. . Additional Symptoms/Keywords KIXREVSVR
Local fix
Set MXT lower than 512 as a temporary workaround.
Problem summary
**************************************************************** * USERS AFFECTED: All CICSPlex SM V4R2M0 Users * **************************************************************** * PROBLEM DESCRIPTION: Your application performs non- * * terminal initiated START for tasks * * which are dynamically routed to target * * regions. Once the tasks in the target * * regions begin running, they might also * * perform a dynamic DPL. Every target * * region receiving work from this work- * * load issues message: * * . * * EYUWG0106E WLM has encountered an * * error while attempting to * * release MAS resources. * * . * * The first time the message is issued in * * each target region, a system dump is * * captured with a dump title beginning * * "EYU0XZSD Dump". * **************************************************************** * RECOMMENDATION: After applying the PTF that resolves this * * APAR, all MASes must be recycled to pick * * up the new code. Note that regions do not * * need to be brought down and restarted at * * the same time. * **************************************************************** When a unit of work is presented to the CPSM Routing Program EYU9XLOP for the first time in a region through a Route Select (RTSELECT) or Route Notify (RTNOTIFY) call in a routing region, or a Route Initiate (RTINIT) call for distributed work arriving in a target region, a WLM Management List Element (WNLE) is ac- quired to anchor storage areas which persist for the duration of the routed work. A pool of WNLEs is allocated at MAS startup with sufficient elements to support the maximum number of tasks which can execute concurrently in a CICS region. Because a failure during the routing process may result in elements not being returned to the free pool, a garbage collection task is scheduled to run if the pool is exhausted, to return elements to the pool if the task that acquired them is no longer active. Message EYUWG0106E is issued if the garbage collection task is unable to return any elements to the pool. The first time the message is issued during execution of a region, a system dump is also taken to help in problem diagnosis. When a distributed task (for example, a non-terminal initi- ated START) is routed to a target region, a WNLE is acquired during RTINIT processing. If the task executes a dynamic DPL request, another WLNE is acquired during RTSELECT processing for the DPL. Since the pool contains 1024 WLNEs, if MAXTASKS is greater than 512, all WLNEs in the pool might be allocated. This will cause the garbage collection task to run. However because all WLNEs are allocated to active tasks, garbage collec- tion will be unable to return any elements to the pool. CPSM will continue to route work by GETMAINing an additional WNLE for each routing task and FREEMAINing it when the task ends, but this adds overhead to the routing process.
Problem conclusion
Of the three blocks anchored by the WNLE, two (Scope Vector and Algorithm Work Area) are not required for distributed work in a target region, and a persistent copy of the third (Trans- action Data Area, TDA) is contained in the DTR COMMAREA, in the DYRUSER area. Therefore a WNLE is not required for distributed work in target regions. Modules EYU0WDIN (WDIN - WDIN Route Initialization and Termination), EYU0WDI7 and EYU0WDI8 (WDI# - CICS Version Specific Init and Term) were modified to address the copy of the TDA in the DYRUSER area so a WNLE need not be acquired. If a task running in a distributed workload issues a dynamic DPL, only one WNLE will be acquired by the task.
Temporary fix
FIX AVAILABLE BY PTF ONLY
Comments
APAR Information
APAR number
PM75803
Reported component name
CICS TS Z/OS V4
Reported component ID
5655S9700
Reported release
70M
Status
CLOSED PER
PE
NoPE
HIPER
NoHIPER
Special Attention
NoSpecatt
Submitted date
2012-10-24
Closed date
2012-12-11
Last modified date
2013-01-02
APAR is sysrouted FROM one or more of the following:
APAR is sysrouted TO one or more of the following:
PM78092 UK90227
Modules/Macros
CJA0WDI7 CJB0WDI7 CJC0WDI7 EYU0WDIN EYU0WDI7 EYU0WDI8
Fix information
Fixed component name
CICS TS Z/OS V4
Fixed component ID
5655S9700
Applicable component levels
R70M PSY UK90227
UP12/12/13 P F212
Fix is available
Select the PTF appropriate for your component level. You will be required to sign in. Distribution on physical media is not available in all countries.
[{"Business Unit":{"code":"BU058","label":"IBM Infrastructure w\/TPS"},"Product":{"code":"SSGMGV","label":"CICS Transaction Server"},"Component":"","ARM Category":[],"Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"4.2","Edition":"","Line of Business":{"code":"LOB35","label":"Mainframe SW"}},{"Business Unit":{"code":"BU054","label":"Systems w\/TPS"},"Product":{"code":"SG19M","label":"APARs - z\/OS environment"},"Component":"","ARM Category":[],"Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"4.2","Edition":"","Line of Business":{"code":"","label":""}}]
Document Information
Modified date:
02 January 2013