APAR status |
Closed as program error.
| Error description
Under certain circumstances, when many new servlets or webapps
re being defined in a short period of time, it is possible for
two or more AdminServers to deadlock, where each AdminServer
is trying to notify the other AdminServers that they need to
regenerate the OSE plugin configuration. As a result, all
AdminServers may stop responding to user requests, and the
entire admin system may freeze-up. Operations attempted from
the AdmiGUI, XMLConfig or wscp hang forever
Defect=89272 Local fixProblem summary
The problem seems to be the PluginCfgGenerator static method
"broadcast" is invoked from a synchronized method. This
method invokes methods on other objects and doing so from
within synchronized code can potentially cause a deadlock
(this is a fundamental rule of concurrent programming).
The potential for a deadlock increases
the more peer nodes (ones that share the same WAS
repository) in an environment. Changed PluginCfgGenerator so
broadcast is not invoked from synchronized code. Problem conclusion
Customer could not test efix as efix was written for 3.5.4
and customer was on older version. APAR closed with no customer
feedback. Temporary fixComments
APAR information | APAR number | PQ48462 | Reported component name | WAS ADVANCED AI | Reported component ID | 5648C8400 | Reported release | 350 | Status | CLOSED PER | PE | NoPE | HIPER | NoHIPER | Submitted date | 2001-05-03 | Closed date | 2001-06-22 | Last modified date | 2001-06-22 |
APAR is sysrouted FROM one or more of the following:
APAR is sysrouted TO one or more of the following:APAR is sysrouted FROM one or more of the following:
Modules/Macros APAR is sysrouted TO one or more of the following:Modules/Macros
|
Fix information |
Fixed component name | WAS ADVANCED AI | Fixed component ID | 5648C8400 |
Applicable component levels | R350 PSY | UP |
|