S878 TSSKERNL+27B30 In Websphere And S09E Abend
Article ID: 190067
Updated On:
Products
Top Secret
Issue/Introduction
The following abend occurred in Top Secret alongside constant Websphere server abends:
TSS9999E CA-TSS SECURITY SVC ABEND S878 IN TSSKERNL+27B30
IEA995I SYMPTOM DUMP OUTPUT 538
SYSTEM COMPLETION CODE=878 REASON CODE=00000010
TIME=11.06.23 SEQ=40319 CPU=0000 ASID=0284
...
CCSR010E TSSRSVCS S878 at 38085950 LMOD TSSMVS2 CSECT TSSRSVCS +010F10
TSS9999E CA-TSS SECURITY SVC ABEND S878 IN TSSRSVCS+10F10
CCSR022I MODULE = TSSRSVCS FMID = CAKOG00 RMID = SO04391
Later, the following abend occurred and Top Secret came down:
IEF196I SYSTEM COMPLETION CODE=09E REASON CODE=00002505
IEF196I TIME=14.08.51 SEQ=60317 CPU=4003 ASID=0028
IEF196I PSW AT TIME OF ERROR 070C2000 B9407F50 ILC 2 INTC 0D
IEF196I NO ACTIVE MODULE FOUND - PRIMARY NOT EQUAL TO HOME
IEF196I NAME=UNKNOWN
TSS9999E CA-TSS SECURITY SVC ABEND S878 IN TSSKERNL+27B30
IEA995I SYMPTOM DUMP OUTPUT 538
SYSTEM COMPLETION CODE=878 REASON CODE=00000010
TIME=11.06.23 SEQ=40319 CPU=0000 ASID=0284
...
CCSR010E TSSRSVCS S878 at 38085950 LMOD TSSMVS2 CSECT TSSRSVCS +010F10
TSS9999E CA-TSS SECURITY SVC ABEND S878 IN TSSRSVCS+10F10
CCSR022I MODULE = TSSRSVCS FMID = CAKOG00 RMID = SO04391
Later, the following abend occurred and Top Secret came down:
IEF196I SYSTEM COMPLETION CODE=09E REASON CODE=00002505
IEF196I TIME=14.08.51 SEQ=60317 CPU=4003 ASID=0028
IEF196I PSW AT TIME OF ERROR 070C2000 B9407F50 ILC 2 INTC 0D
IEF196I NO ACTIVE MODULE FOUND - PRIMARY NOT EQUAL TO HOME
IEF196I NAME=UNKNOWN
Environment
Release : 16.0
Component : CA Top Secret for z/OS
Resolution
Top Secret was a victim in both problems.
** The S878 abends in Websphere were a result of a combination of:
1) A large number of threads.
2) 2 hour (default) expiration time for the Websphere token.
When the servers are first started, an authentication is done and a token created for the user. All users are using the same token and the token expires (by default) after 2 hours. After the expiration, the user has to be re-authorized and the token re-built. Until the first user completes this process again and the token is cached, all users go through the authorization and token build. The large number of threads allowed causes a large amount of processing to be done each time the token expires and this uses up a lot of storage in the region. The process occurs every 2 hours. When the S878 abend occurred, all the storage had been used up. Top Secret was a victim in this case because there was no storage available for Top Secret to process the request, so the S878 abend was issued.
For this problem:
1) Reduce the number of threads to limit the number of simultaneous users going through the signon process.
2) Increase the expiration value Websphere to more than 2 hours.
** For the S09E, IBM said the problem is a systrace abend that is the result of a serialization problem. The trigger is a simultaneous occurrence of a mem term and a task termination which causes the owner to be lost. (In this case, Top Secret was the owner.) IBM is going to look into writing an APAR to correct this.
** The S878 abends in Websphere were a result of a combination of:
1) A large number of threads.
2) 2 hour (default) expiration time for the Websphere token.
When the servers are first started, an authentication is done and a token created for the user. All users are using the same token and the token expires (by default) after 2 hours. After the expiration, the user has to be re-authorized and the token re-built. Until the first user completes this process again and the token is cached, all users go through the authorization and token build. The large number of threads allowed causes a large amount of processing to be done each time the token expires and this uses up a lot of storage in the region. The process occurs every 2 hours. When the S878 abend occurred, all the storage had been used up. Top Secret was a victim in this case because there was no storage available for Top Secret to process the request, so the S878 abend was issued.
For this problem:
1) Reduce the number of threads to limit the number of simultaneous users going through the signon process.
2) Increase the expiration value Websphere to more than 2 hours.
** For the S09E, IBM said the problem is a systrace abend that is the result of a serialization problem. The trigger is a simultaneous occurrence of a mem term and a task termination which causes the owner to be lost. (In this case, Top Secret was the owner.) IBM is going to look into writing an APAR to correct this.
Feedback
Yes
No
Powered by
