jBASE Bus error
JOSE L MARTINEZ-AVIAL
We are having some troubles with T24 (R7.00) and Jbase 5.0.14. This issue seems to be Jbase, but it only happens from Globus. I'm trying to input a record, and I am getting the following error:
jBASE: Bus error. Aborting
<Your 'TELNET' connection has terminated>
I received the same error from Globus Clasic, and from Globus Desktop. It's related to Globus, but since it seems to be a Jbase error, I didn't put the T24 in the subject. Any one has seen this error? any advices about where the problem may be?
jsh miadev02 ~ -->jdiag
jdiag - jBASE diagnostic '$Revision: 1.15 $'
System Information
==================
System : HP-UX miaux02 B.11.11.U 9000/800
UNIX User : miadev02 (uid 107, euid 107)
Tty name : /dev/pts/ts
Time : Tue Nov 17 13:56:33 2009
Environment
===========
JBCPORTNO : Not Set
JBCRELEASEDIR : '/apps/jbase5.0.14'
JBCGLOBALDIR : '/apps/jbase5.0.14'
JBCDATADIR : '/ve8/miadev02/bnk/bnk.run/../bnk.data'
HOME : '/ve8/miadev02/bnk/bnk.run'
JEDIFILEPATH : '/ve8/miadev02/bnk/bnk.run'
JEDIFILENAME_MD : '/ve8/miadev02/bnk/bnk.run/VOC'
JEDIFILENAME_SYSTEM : '/apps/jbase5.0.14/src/SYSTEM'
SYSTEM File is (DICT) : '/apps/jbase5.0.14/src/SYSTEM]D'
RELEASE Information : Major 5.0 , Minor 14 , Patch 0287 (Change 73758)
Spooler dir (JBCSPOOLERDIR) : '/ve8/miadev02/bnk/bnk.run/jspooler'
JBCEMULATE : 'prime'
WARNING: Cannot access Executable path '/ve8/miadev02/bnk/bnk.run/globuspatchbin', error 2
WARNING: Cannot access Library path '/apps/jbase5.0.14/jdk/jre/lib', error 2
Object path (JBCOBJECTLIST) : '/ve8/miadev02/bnk/bnk.run/globuspatchlib:/ve8/miadev02/bnk/bnk.run/lib:/ve8/miadev02/bnk/bnk.run/glo'
jBASE Compiler Run-time : '/apps/jbase5.0.14/config/system.properties'
Program dir (JBCDEV_BIN) : '/ve8/miadev02/bnk/bnk.run/bin'
Subroutine dir (JBCDEV_LIB) : '/ve8/miadev02/bnk/bnk.run/lib'
Max open files : 2048
Jim Idle
José,
Congratulations on thinking about your post for more than two seconds and supplying as much information as you could.
Bus errors occur because some program has attempted to access an area of memory that was not allocated to it. In jBASE, because there are so many applications running the same code all the time, it is rarely caused by a problem inherent to jBASE or the jBASE runtime. The most common cause for this is a corrupt file, which causes jBASE to try and follow links to memory addresses that it will not actually own. This is made worse because HPUX is generally a pretty weak version of UNIX to be honest (I think I may have intimated my views on HPUX before now ;-).
So, you should get everyone off the system (that is an important piece) and use the jcheck program to see if you can find out which, if the assumption here is correct, file is corrupt. If you do not find any corrupt files in your data, then the issue might be something in the jBASE tmp file set. You can usually fix that by deleting everything in the tmp directory (/apps/jbase5.0.14/tmp) when everyone is off the system. Many people do this routinely at operating system boot time. If your file data is very large, then you might want to try this first.
If, you fail to identify any corrupt data files, then the issue may really be some inherent problem and will take more tracking down. You will need to use a program like tusc, which I think works on HPUX 11.11, and failing that then the gdb program. All you need do is run the login process using that program and repeat the steps to reproduce the issue. When you get to the bus error, gdb will stop and give you some information and tusc will give similar information but possibly more useful. If your system is set up to allow core dumps (the process memory is dumped to a file called core), then you might not need to run
Next, if you have a corrupt file, then you need to work out why, so before trying to fix or restore it, make a copy of it to give to TEMENOS support. I have to tell you though that biggest source of file corruption is in fact human error. So you should be using journaling to get an ongoing backup of your database. The ways that this can happen are discussed at length in past posts, just look for “kill corruption” in http://jbase.markmail.org . But the usual culprits are people having access to the kill -9 command, and turning off the system without shutting down properly (which is usually OK but not always).
Jim
--
Please read the posting guidelines at: http://groups.google.com/group/jBASE/web/Posting%20Guidelines
IMPORTANT: Type T24: at the start of the subject line for questions specific to
Globus/T24
To post, send email to jB...@googlegroups.com
To unsubscribe, send email to jBASE-un...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/jBASE?hl=en