Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.
Bug 161220 - Linux (x86) Agent Controller does not start with IBM V5 SR3 JRE.
Summary: Linux (x86) Agent Controller does not start with IBM V5 SR3 JRE.
Status: CLOSED FIXED
Alias: None
Product: z_Archived
Classification: Eclipse Foundation
Component: TPTP (show other bugs)
Version: unspecified   Edit
Hardware: PC Linux
: P1 critical (vote)
Target Milestone: ---   Edit
Assignee: Samson Wai CLA
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2006-10-17 08:38 EDT by Paul Slauenwhite CLA
Modified: 2016-05-05 10:51 EDT (History)
1 user (show)

See Also:


Attachments
Patch file (28.98 KB, patch)
2007-02-07 13:53 EST, Samson Wai CLA
no flags Details | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description Paul Slauenwhite CLA 2006-10-17 08:38:57 EDT
Linux (x86) Agent Controller does not start with IBM V5 SR3 JRE.

Using the TPTP-4.3.0-200610160100 build, when attempting to start the new Agent Controller on the
Linux (Linux paules 2.4.27-2-386 #1 Mon May 16 16:47:51 JST 2005 i686
GNU/Linux), the following error is displayed to the console and no service log is generated:

Cannot find VM.

Using the IBM V5 SR2 JRE with the same configuration works.
Comment 1 Samson Wai CLA 2006-10-23 12:25:31 EDT
I think this is somewhat related to how the jvm libraries are built. There was a problem when testing pre-SR3 is that libjvm.so is dependent on other libraries under another directory - mostly in the parent dir of libjvm.so. This symptom is also found on Windows/IA32 as well during pre-SR3. I doubt the JVM team had this issue resolved in SR3 and a recommended workaround is to have our product setting up the LD_LIBRARY_PATH properly when creating the JVM using the libjvm.so library. This should be done in the RAStart.sh script if not yet done so.

Of course I am not ruling out a problem in the RAStart.sh script itself.
Comment 2 Karla Callaghan CLA 2006-10-24 13:16:57 EDT
I compared the RAStart scripts for RAC and AC in how they set the LD_LIBRARY_PATH and they look the same.  So if it is a script problem, this will have to be resolved in the RAC pkg for 4.3.
Comment 3 Thayaparan Shanmugaratnam CLA 2006-10-25 12:19:05 EDT
This problem is not re-producible with the TPTP-4.3.0-200610230100.  The TPTP-4.3.0-200610160100 driver is built using the new AC, while the 1023 driver is built using the RAC.  
Since it has been agreed upon to ship TPTP 4.3 with the RAC, we can re-target this bug to a future release.
Comment 4 Samson Wai CLA 2006-10-26 12:32:17 EDT
This cannot be reproduced using TPTP 4.3 RAC so retarget to 4.4 where the new AC is going replace RAC on Linux/IA32.
Comment 5 Samson Wai CLA 2007-01-31 16:09:41 EST
Set priority to P1 for 4.4 plan closure.
Comment 6 Samson Wai CLA 2007-02-01 15:54:52 EST
The script for the new AC is not correct when parsing serviceconfig.xml for the jvm library location. The RAC script looks for the attribute "jvm=..." while the new AC should look for the element <Jvm>...</Jvm> instead. Fixing the script should make AC starts.
Comment 7 Samson Wai CLA 2007-02-05 15:18:25 EST
There are a few things that stop the AC from coming up:
1. The name of the keystore does not match - it is case sensitive!
2. There is no ChkPass utiliy for AC - need to migrate from the RAC
3. RAStart.sh cannot determine the location of the JVM_LIB since the RAC looks for the "jvm" attribute while the AC should look for the <Jvm> text
4. Also on my system the libcrypt.so library is missing - it complains about -lcrypt at runtime

Need to fix all of the above.
Comment 8 Samson Wai CLA 2007-02-07 13:53:50 EST
Created attachment 58472 [details]
Patch file

This patch may solve the startup problem.
Comment 9 Samson Wai CLA 2007-02-07 14:09:56 EST
Fix checked in 2007/02/07 14:09 EST.
Comment 10 Samson Wai CLA 2007-02-07 14:11:46 EST
Bug 173319 is opened to track another problem which is not related to the startup of the AC. The new shared memory transport layer fails to come up and need to be disabled in order to startup the AC.
Comment 11 Paul Slauenwhite CLA 2007-06-27 14:58:49 EDT
Verified in TPTP-4.4.0-200706140100C.