There are 2 reasons to get MQRC 2071. Either there is not enough memory/disk space for the user running the process ( ulimit settings). Or someone tried to put an empty message on a queue. There is a fix available on MQ for empty messages.
Will ad more detail later.
Tuesday, September 28, 2010
Friday, September 3, 2010
MQ v7 client jars upgrade on WAS 6, 6.1 or 7.0
We had a lot of issues here at my customer when we tried upgrading the client jars for MQ from v6 to v7.0.1.2.
You know how MQ client jars though packaged with WAS can be externalized and you could have WAS 6.0.x updated to use MQ v7 client jars.
Problems :
1. Unable to put messages on an MQ Queue.
2. Unable to retrieve messages via an MDB.
3. MQ v7 client jars added a new behavior where if no ID was specified, it would use the WAS process running user ID as the ID for authentication on the MQ Qmgr. With MQ V6 client jars a blank ID was passed, if no ID was specified. MQ QMgr runs under an MCA user ID and if a blank was encountered it would authenticate using the MCA user ID and it would work. However , since MQ v7 client jars specified a fixed ID, the WAS running process user ID need to be in the mqm group for it to be authenticated on the Qmgr. This was a problem.
4. Issues serializing a JMS Message originated from MQ v6 Qmgr and delivered via the MQ v7 client jars to WAS.
Easy fix:
1. Most customers use a component managed auth alias configured on their MQ Queue Connection factories, but this customer viewed component managed auth aliases as a security hole and wanted it to work with container managed auth aliases only.
2. Include the WAS running process User id in the mqm group.
Resolution:
IZ72486 - A new APAR was created that rolled back the blank ID behavior on the MQ v7 client jars. So now a blank ID would stay blank.
( this will be included after 7.0.1.3)
PMR 05950,122,000 - IZ81411 - A new APAR that allowed some of the MQ v6 custom headers in the JMSObject to be serialized in the MQ v7 client jars so that they can be passed over the wire. ( this will be included later, maybe after 7.0.1.4).
The best and easiest way to get the fixes was to go by the MQ client jar version. Fixers asked for WAS were getting garbled up because support tended to match a WAS version with the appropriate MQ version as I mentioned above that the MQ client jars are bundled along with the product.
You know how MQ client jars though packaged with WAS can be externalized and you could have WAS 6.0.x updated to use MQ v7 client jars.
Problems :
1. Unable to put messages on an MQ Queue.
2. Unable to retrieve messages via an MDB.
3. MQ v7 client jars added a new behavior where if no ID was specified, it would use the WAS process running user ID as the ID for authentication on the MQ Qmgr. With MQ V6 client jars a blank ID was passed, if no ID was specified. MQ QMgr runs under an MCA user ID and if a blank was encountered it would authenticate using the MCA user ID and it would work. However , since MQ v7 client jars specified a fixed ID, the WAS running process user ID need to be in the mqm group for it to be authenticated on the Qmgr. This was a problem.
4. Issues serializing a JMS Message originated from MQ v6 Qmgr and delivered via the MQ v7 client jars to WAS.
Easy fix:
1. Most customers use a component managed auth alias configured on their MQ Queue Connection factories, but this customer viewed component managed auth aliases as a security hole and wanted it to work with container managed auth aliases only.
2. Include the WAS running process User id in the mqm group.
Resolution:
IZ72486 - A new APAR was created that rolled back the blank ID behavior on the MQ v7 client jars. So now a blank ID would stay blank.
( this will be included after 7.0.1.3)
PMR 05950,122,000 - IZ81411 - A new APAR that allowed some of the MQ v6 custom headers in the JMSObject to be serialized in the MQ v7 client jars so that they can be passed over the wire. ( this will be included later, maybe after 7.0.1.4).
The best and easiest way to get the fixes was to go by the MQ client jar version. Fixers asked for WAS were getting garbled up because support tended to match a WAS version with the appropriate MQ version as I mentioned above that the MQ client jars are bundled along with the product.
Wednesday, June 2, 2010
Container vs Component managed authentication aliases
In WebSphere JAAS aliases can be defined for 3rd party resources like database/ MQ/ JMS etc. There is a property res-auth in the deployment descriptor that defines which auth alias to use. There are two kinds of res-auth container or application. For container we use container managed authentication aliases. For application we use component managed authentication aliases. The default if no res-auth is specified is application.
For MQ Listener Ports, there is no place to specify a res-auth and hence it defaults to the J2C default res-auth which is application and hence if security credentials need to be passed on to MQ from WAS via a QCF used in a listener port a component managed authentication alias is required.
For MQ Listener Ports, there is no place to specify a res-auth and hence it defaults to the J2C default res-auth which is application and hence if security credentials need to be passed on to MQ from WAS via a QCF used in a listener port a component managed authentication alias is required.
Tuesday, April 27, 2010
Windows 7 has a new security feature. Due to this feature we get "channel binding" checksum failed errors on WAS V 6.0, 6.1, 7.0 when negotiating a kerberos token with the TAI. IZ64838 is the jdk bug
PM10213 is the JDK ifix for IZ64838.
That fix is available for IBM internals at
autoJDK.rtp.raleigh.ibm.com/autoJDK
PM10213 is the JDK ifix for IZ64838.
That fix is available for IBM internals at
autoJDK.rtp.raleigh.ibm.com/autoJDK
Monday, April 12, 2010
| New WS Application Server APARs Encountered by Customers for 03/19/2010 - 03/25/2010 | ||||
APAR |
Open Date |
Product | Version | Description |
| IZ73240 |
03/23/2010 | BUS PRC CHOREOGMP | 620 | STATES NOT REFRESHED IN EXPLORER LISTS USING QUERY TABLES |
| IZ73243 |
03/23/2010 | BUS PRC CHOREOGMP | 620 | WHEN DELETING TASK INSTANCES SOME ENTRIES IN THE RETRIVED_USER_TTABLE MIGHT NOT BE DELETED. |
| IZ72262 |
03/25/2010 | BUS PRC CHOREOGMP | 612 | INCOMPLETE CLEANUP WHEN BPEL WHILE LOOPS ARE USED. |
| IZ73386 |
03/25/2010 | BUS PRC CHOREOGMP | 612 | NULLPOINTEREXCEPTION IN REST API WHEN USING EVENT HANDLER IN A BUSINESS PROCESS. |
| IZ73242 |
03/23/2010 | BUS PRC CHOREOGMP | 612 | WHEN DELETING TASK INSTANCES SOME ENTRIES IN THE RETRIVED_USER_TTABLE MIGHT NOT BE DELETED. |
| IZ73241 |
03/23/2010 | BUS PRC CHOREOGMP | 610 | WHEN DELETING TASK INSTANCES SOME ENTRIES IN THE RETRIVED_USER_TTABLE MIGHT NOT BE DELETED. |
| IZ73232 |
03/25/2010 | J9 COMMON CODE | 600 | SIGNAL CHAINING FUNCTIONALITY ISSUE |
| IZ73271 | 03/25/2010 | J9 COMMON CODE | 600 | ADD TRACEPOINTS TO CLASS LOADING CONSTRAINT CHECKS |
| IZ73273 | 03/25/2010 | J9 COMMON CODE | 600 | VMS RUNNING AS ROOT CAN NOT BE ATTACHED TO OTHER USERS' VMS |
| IZ73097 |
03/23/2010 | J9 COMMON CODE | 600 | RACF MESSAGES WHILE ACCESSING SHARECLASS CACHE |
| IZ73156 | 03/23/2010 | J9 COMMON CODE | 600 | WIN64 JVM ALLOCATES THE JAVA HEAP BELOW THE 4 GB MARK LEADING TO OOME FAILURES |
| IZ73141 | 03/23/2010 | JAVA 5 CLASS LIBS | 500 | UPDATE TO VERSION TZDATA2010F OF THE OLSON DATABASE |
| IZ73214 |
03/25/2010 | JAVA 5 J9 COMMON | 500 | SIGNAL CHAINING FUNCTIONALITY ISSUE |
| IZ73272 | 03/25/2010 | JAVA 5 J9 COMMON | 500 | ADD TRACEPOINTS TO CLASS LOADING CONSTRAINT CHECKS |
| IZ73277 | 03/25/2010 | JAVA 5 J9 COMMON | 500 | VMS RUNNING AS ROOT CAN NOT BE ATTACHED TO OTHER USERS' VMS |
| IZ73096 |
03/23/2010 | JAVA 5 J9 COMMON | 500 | RACF MESSAGES WHILE ACCESSING SHARECLASS CACHE |
| IZ73098 | 03/23/2010 | JAVA 5 JIT | 500 | APPLICATION ERROR CAUSED BY INCORRECT OPTIMIZATION |
| IZ73212 |
03/24/2010 | JAVA 5 ORB | 500 | WCHAR DO NOT WORK IN COMPLEX DATA TYPES. UPON MARSHALLING IT THROWS MARSHAL EXCEPTION: TCKIND._TK_SEQUENCE EXCEPTION |
| IZ72948 | 03/20/2010 | JAVA CLASS LIBS | 600 | PROBLEMS WHILE USING JOBNAME FOR PRINT JOBS ON Z/OS |
| PM10432 | 03/22/2010 | JAVA(1.3/1.4 COMM | 42A | UPDATE TO VERSION TZDATA2010F OF THE OLSON DATABASE |
| IX90050 | 03/25/2010 | ORB | 600 | RMIC GENERATES IDL CODE WITH DUPLICATE ATTRIBUTES |
| IX90049 | 03/19/2010 | ORB | 600 | *ORB*: Large ClientDelegate instances leading to heap exhaustion |
| PM06411 | 03/24/2010 | WAS COMMUNITY EDIT |
210 | REDEPLOY OPTION FOR FARMING |
| PM10658 | 03/24/2010 | WAS HTTP SERVER 2.0 | 10A | IBM HTTP SERVER 2.0.47 CUMULATIVE INTERIM FIX |
| PM10316 |
03/22/2010 | WAS IHS 64BIT PLUGI | 700 | HANG IN WEBSPHERE APPSERVER AFTER CLIENT CLOSES BROWSER IN THE MIDDLE OF A RESPONSE |
| PM10380 |
03/23/2010 | WAS IHS 64BIT PLUGI | 610 | PLUGIN RETURNS RESPONSE CODE 500 WHEN USING URL REWRITE FOR SESSION AFFINITY. $WSFO SET PREMATURE BASED ON JSESSIONID |
| PM09733 |
03/21/2010 | WAS NETWRK DEPLOY 5 | 510 | WEBSPHERE APPLICATION SERVER MAY ALLOW USER REGISTRY ACCESS BEFORE INITIALIZATION COMPLETES |
| PM07769 |
03/22/2010 | WAS XD/EXTREME SCAL | 60A | DEFAULTCOOOKIESETTINGS NAME PROPERTY IS NOT PROPAGATED TO ALL OFTHE MEMBERS IN THE DYNAMIC CLUSTER WHEN SET AT THE TEMPLATE. |
| PM10631 | 03/24/2010 | WEBS APP SERV ND | 700 | MEMORY CONSUMPTION/GARBAGE COLLECTION ISSUE IN THE UNIFIED EL PACKAGED IN WEBSPHERE APPLICATION SERVER V 7.0.0.5 |
| PM10450 |
03/22/2010 | WEBS APP SERV ND | 700 | SWS3173E: ERROR: DID NOT UNDERSTAND "MUSTUNDERSTAND" WITH 'MUSTUNDERSTAND=1' |
| PM10321 | 03/21/2010 | WEBS APP SERV ND | 700 | OBPL0006E ISSUED IF OBJECTPOOL LOOKUP IS DONE IN STARTUPBEAN |
| PM10353 | 03/21/2010 | WEBS APP SERV ND | 700 | WEBSPHERE PLUGIN WILL RETRY POSTS WHEN POSTBUFFSIZE IS SET TO 0 |
| PM10442 |
03/21/2010 | WEBS APP SERV ND | 700 | AHE BSTRACT: SAML FEATURE PACK DOES NOT HAVE CLOCK SKEW SUPPORT ON THE GENERATOR SIDE. |
| PM10278 |
03/21/2010 | WEBS APP SERV ND | 61S | WSAS APAR FOR DELIVERY OF SOLARIS JAVA BUILDS FOR -XX:+PRINTCLASSHISTOGRAM TESTING. |
| PM10332 |
03/19/2010 | WEBS APP SERV ND | 61S | PREVENT SCHEDULING OF DUPLICATE ODC TREE REQUESTS |
| PM10331 | 03/21/2010 | WEBS APP SERV ND | 61A | AFTER APPLYING 6.1.0.29 CONNECTIONS TO WMQ FAIL WITH MQRC 2035 |
| PM10424 |
03/21/2010 | WEBS APP SERV ND | 61A | SLOW MEMORY BUILD UP IN COM/IBM/WS/SECURITY/UTIL/ |
| PM10568 |
03/23/2010 | WEBS APP SERV ND | 60A | SHIP SDK APAR IZ63865 AS WSAS IFIX - JSSE2 SOCKET OBJECTS LEAKING IN THE HEAP |
| PM10745 |
03/25/2010 | WEBS LOAD BALANC | 700 | LOAD BALANCER IPV4 AND IPV6-IF THE CROSSPORT OPTION OF DSCONTROLIS USED WHEN THE SELECTIONALGORITHM THE SYSTEM MAY CRASH |
| PM10452 | 03/22/2010 | WEBSERVIC FEATURPAK | 610 | While running a JAX-WS application an IllegalAnnotationsException may occur |
| PM10871 |
03/25/2010 | WEBSPHERE APP SERV |
700 | A message sent from a JAX-RPC application may have an element named arg0, which does match the element defined in the WSDL |
| PM10292 | 03/23/2010 | WEBSPHERE APP SERV |
700 | A JAX-WS CLIENT MAY FAIL WITH "IOEXCEPTION: ASYNC IO OPERATION FAILED" ERROR IF ITS HOST HEADER CONTAINS A PORT |
| PM10472 |
03/23/2010 | WEBSPHERE APP SERV |
700 | SHIP SDK APAR IZ66798 FOR WSAS IFIX |
| PM10447 |
03/22/2010 | WEBSPHERE APP SERV |
700 | USING DB2 JCC DRIVER,ON FATAL ERROR NOTIFICATION PREPARED STATEMENT CACHE CLEARS CHILD WRAPPERS.THIS IS INCORRECT BEHAVIOR |
| PM10352 | 03/21/2010 | WEBSPHERE APP SERV |
61Z | EMBEDDED SHARED RESOURCES ARE ONLY ADDED TO ONE SERVER WHEN TARGETING MORE THAN ONE SERVER |
| PM10362 |
03/22/2010 | WEBSPHERE APP SERV |
61A | On 6.1.0.27,DBCS URL requestfails w("404 Error") for korean fileon AIX en_US locale |
| PM10161 |
03/19/2010 | WEBSPHERE APP SERV |
61A | WLM NULLPOINTER IN CLUSTER MBEAN |
| PM10622 | 03/24/2010 | WEBSPHERE APP SERV |
60W | HEXACHARACTER LOGIN ISSUE |
| PM10657 |
03/24/2010 | WEBSPHERE APP SERV |
60W | HEXACHARACTER LOGIN ISSUE |
| PM10485 |
03/24/2010 | WEBSPHERE FOR Z/OS |
700 | POSSIBLE REUSASID=YES ISSUES INTERACTING WITH OTHER PRODUCTS AFTER RUNNING UPDATEZOSSTARTARGS.PY |
| PM10487 |
03/22/2010 | WEBSPHERE FOR Z/OS |
700 | WITH RUNTIME PROVISIONING ENABLED, AXIS2 SERVICE FAILS TO START |
| PM10273 |
03/19/2010 | WEBSPHERE FOR Z/OS |
700 | CONFIG FILE NAME TOO LONG FOR WEBSERVER CAN CAUSE AN ABEND |
| PM10291 |
03/19/2010 | WEBSPHERE FOR Z/OS |
700 | CONTROLLER REGION TERMINATES WITH MINOR CODE C9C21A4D |
| PM10308 |
03/19/2010 | WEBSPHERE FOR Z/OS |
700 | UPDATING THE WMQ RESOURCE ADAPTER CAUSES BINARIES DELETETION IN INSTALLEDCONNECTORS DIR., THEREBY NOT WORKING FOR OTHER PROFILES |
| PM10317 |
03/19/2010 | WEBSPHERE FOR Z/OS |
700 | METHOD NAMES IN MESSAGE BBOO0325W ARE UNPRINTABLE GARBAGE CHARACTERS. |
| PM10711 |
03/25/2010 | WEBSPHERE FOR Z/OS |
610 | JAVA JIT COMPILATION THREAD NOT RELEASING VM CLASS TABLE LOCK CAUSING APPLICATION TO HANG. |
| PM10302 |
03/19/2010 | WEBSPHERE FOR Z/OS |
610 | METHOD NAMES IN MESSAGE BBOO0325W ARE UNPRINTABLE GARBAGE CHARACTERS. |
| PM10297 | 03/19/2010 | WEBSPHERE FOR Z/OS |
601 | FFDC SUBSYSTEM SHUTS DOWN BUT DOES NOT REPORT IT. |
| PM09745 | 03/19/2010 | XD COMPUTE GRID | 610 | DUPLICATE SUBJOBS ARE CREATED. |
| PM09395 |
03/22/2010 | XD VIRTUAL ENTERPRI | 610 | UPDATING APPLICATIONS WITH DIFFERENT EDITIONS AND DIFFERENT WEBMODULE NAMES RESULTS IN CORRUPTED WORKCLASSES. |
| PM09676 | 03/22/2010 | XD VIRTUAL ENTERPRI | 610 | The middleware agent cannot be federated when security is enabled. |
Subscribe to:
Comments (Atom)