Community
Participate
Working Groups
This problem occurs on a Windows EM64T server. At termintaion of a remote profiled app, sometimes the ACServer abends (0xC05 at hcclsm.dll+0x1da1) and sometimes it continues but the last few console sends fail to appear at the client (although they are logged in the servicelog.log). It is intermittent so far (about once every two or three trials). This failure occurred with the TPTP-4.3.0-200611070100 driver and the Sun 1.5.0_06 jdk. It occurs with or without security enabled. The steps to reproduce are: 1. Install RAC on an EM64T server. 2. Start Workbench and change to Profiling and Logging perspective 3. Profile the InOut testcase (which uses console) as an External Java App. Monitor Execution Statistics.
Created attachment 53768 [details] Windows error log from abend.
Created attachment 53769 [details] AC service log from abend (with format=Simple, level=Debug)
Created attachment 53770 [details] Test class used.
Created attachment 53771 [details] Java for test class used.
Becuase this is intermittent, I am leaving the sev at major for now.
The abend never occurs for workbench-attached profiling. It only occurss for workbench-launched profiling. This is further evidence that the problem is related to the handling of the console (which totally bypasses the AC and workbench in workbench-attached scenarios).
The failures (abend and/or lost console output) are definitely more intermittent if the ACserver is installed as a Windows service. In that case, the failures occur once per approximatley 10 trials.
I just reviewed your servicelog.log file... It looks like the AC is collecting the console information. For some reason when the process is terminating we are not waiting for the final flush of the console before we close the console connection. I have seen issues like this in the past but I thought we had resolved all of them.
I have been unable to get this to fail using the automated test suite. I have run the ConsoleTest for > 10 iterations and it works each time.
Ok... I finally read all the details of this bug... and started profiling external java apps. I am not seeting console data being lost but I am seeing the AC crash at the address you specify. It seems intermittent... maybe 1 time in 5... but still this is bad... this is the only platform I am seeing this. The particular error that I am seeing seems unrelated to the console... console does not use shared memory... When the crash does occur I am seeing a Hot Spot java dump... # An unexpected error has been detected by HotSpot Virtual Machine: # # EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x0000000001f81da1, pid=820, tid=504 # # Java VM: Java HotSpot(TM) 64-Bit Server VM (1.5.0_06-b05 mixed mode) # Problematic frame: # C [hcclsm.dll+0x1da1] # --------------- T H R E A D ---------------
The AC crash occurs in the shared memory library, which would explain why this problem is platform-specific (shared memory is EM64T specific and the console impl is not). The console app may set things up for failure or it may be conincident that the failure shows up with it. This is not something we expect can be found/fixed quickly. Given the isolation to one platform and that it is an intermitent failure, defer to 4.4.
*** Bug 165374 has been marked as a duplicate of this bug. ***
Added effort estimate: 5 days
Hi Bob, Is this bug still reproducible? I can't reproduce it using TPTP 4.4i2 build (runtime and Agent Controller).
I cannot reproduce this case or the duplicate referenced above with 4.4.0i2. One difference though is in 4.4 I am testing using a remote workbench on Windows XP because there is not yet an Eclipse 3.3 for EM64T windows. In 4.3 I would have been using the workbench locally.
Resolving as "works for me"
As of TPTP 4.6.0, TPTP is in maintenance mode and focusing on improving quality by resolving relevant enhancements/defects and increasing test coverage through test creation, automation, Build Verification Tests (BVTs), and expanded run-time execution. As part of the TPTP Bugzilla housecleaning process (see http://wiki.eclipse.org/Bugzilla_Housecleaning_Processes), this enhancement/defect is verified/closed by the Project Lead since this enhancement/defect has been resolved and unverified for more than 1 year and considered to be fixed. If this enhancement/defect is still unresolved and reproducible in the latest TPTP release (http://www.eclipse.org/tptp/home/downloads/), please re-open.