wainwrr
Well Known Member
Xe SP23_S1 WebSphere 6 multiple cluster members Lots of Web clients
It's a quiet day in Gotham city when suddenly CNCman notices that the JAS log file on one cluster has run up to 10Gb. A quick tail of the log files reveals that the cause is a constantly repeating failure of an insert into an HR tail - failing because of an attempted primary key violation and the 11Gb log file is still growing.
Tracing back from the log file the virtual client is identified in SAW as belonging to a particular user. Visisting their workstation to find out why we've got a 12Gb log file, we see on screen the cogs spinning. With a 13Gb log file we kill the IE session on the workstation and log the user out of JDE. However, when we reach 14Gb, looking in saw the user still appears logged in with all the virtual clients still present.
Round about this time the web server runs out of dsik and effectively halts the sessions. So we stop the cluster member, delete the 16Gb log file, restart and everything is well again.
So...........
What could we have done - apart from never letting that App near Live in the first place?
How with WebSphere, SAW etc. can we kill a user individually or preferably a single virtual client and not have to resort to taking out the whole cluster member and the 20 attached users?
It's a quiet day in Gotham city when suddenly CNCman notices that the JAS log file on one cluster has run up to 10Gb. A quick tail of the log files reveals that the cause is a constantly repeating failure of an insert into an HR tail - failing because of an attempted primary key violation and the 11Gb log file is still growing.
Tracing back from the log file the virtual client is identified in SAW as belonging to a particular user. Visisting their workstation to find out why we've got a 12Gb log file, we see on screen the cogs spinning. With a 13Gb log file we kill the IE session on the workstation and log the user out of JDE. However, when we reach 14Gb, looking in saw the user still appears logged in with all the virtual clients still present.
Round about this time the web server runs out of dsik and effectively halts the sessions. So we stop the cluster member, delete the 16Gb log file, restart and everything is well again.
So...........
What could we have done - apart from never letting that App near Live in the first place?
How with WebSphere, SAW etc. can we kill a user individually or preferably a single virtual client and not have to resort to taking out the whole cluster member and the 20 attached users?