OK, this really was a problem. I've uploaded a new version of
SDK 1.2.2/A.22.05 to Jazz that incorporates a fix for the problem.
(If you're interested, the problem is that the setLength() call
was trying to set file length using a 64-bit field, and the
underlying MPE call only takes a 32-bit parameter).
The java version number didn't change with this fix.
You should see a more reasonable EOF on the logfile now (starting out
at 8)
Mike
Greg Chaplin ([log in to unmask]) wrote:
: Mike (& any others),
: I still can't get rmid to work.
: I was using Java 1.2.2 A.22.02, so I installed the very latest version from Jazz, JDK122 A.22.05, and tried to run rmid again with debug switched on. It died again :-( with the following message:
: /AUSSST/WEBSITE/agent shell/iX : sh rmid.sh
: log.debug: recover()
: log.debug: recovering from /AUSSST/WEBSITE/rmid/Snapshot.1
: log.debug: reading updates from /AUSSST/WEBSITE/rmid/Logfile.1
: log.debug: last update incomplete
: log.debug: recovered updates: 0
: Activation.main: an exception occurred: System call error
: java.io.IOException: System call error
: at java.io.RandomAccessFile.setLength(Native Method)
: at sun.rmi.log.ReliableLog.recover(ReliableLog.java:183)
: at sun.rmi.server.Activation.main(Compiled Code)
: Note that the error messages are different under the new version.
: The shell script is very simple (but the same thing happens when I execute each command directly):
: /AUSSST/WEBSITE/agent shell/iX : cat rmid.sh
: export CLASSPATH=$CLASSPATH:$WEBSITECLASSPATH
: rmid -J-Djava.security.policy=$WebSitePolicyFile -J-Dsun.rmi.log.debug="true" -log /AUSSST/WEBSITE/rmid &; echo $! >/tmp/rmid.pid; wait
: The directory contains the following files (note the size of Logfile.1, and it appears to be empty?)
: /AUSSST/WEBSITE/agent shell/iX : ls -l ../rmid
: total 2263082
: -rw-rw---- 1 MGR.AUSSST AUSSST 1158696736 Jan 31 14:30 Logfile.1
: -rw-rw---- 1 MGR.AUSSST AUSSST 237 Jan 31 14:30 Snapshot.1
: -rw-rw---- 1 MGR.AUSSST AUSSST 4 Jan 31 14:30 Version_Number
: ------------------------------------------------------------------------------------------
: Greg Chaplin
: www.unisuper.com.au
: [log in to unmask]
: UniSuper
: Level 37, 385 Bourke Street
: Melbourne VIC 3000,
: Australia.
: Phone: 61 3 9691 4145
: Fax: 61 3 9691 4141
: >>> Mike Yawn <[log in to unmask]> 31/01/01 4:47:22 >>>
: Hi,
: I haven't used rmid myself, so I'm certainly no expert on what
: might go wrong with it.
: But looking at the tool docs, it appears a security policy file
: might be required. See the page at
: http://java.sun/com/j2se/1.3/docs/tooldocs/solaris/rmid.html
: Although the doc refers to Solaris, we're using exactly the
: same implementation of rmid for MPE/iX.
: You might also turn on a debug switch that's in the code to see
: if it gives any additional information about the failure:
: rmid -J-Dsun.rmi.log.debug="true"
: There should be a 'log' subdirectory created when you run
: rmid, is it being created, and does it have any files in it?
: The IOException may indicate inability to create a directory
: or to create files in it.
: If none of these help, send me any additional stack traces
: that you get (if any are different than the one shown). A
: 'System call error' usually indicates a POSIX call within
: the JVM failed, and it may be something that needs a fix
: within the VM itself if the above suggestions don't get
: around it.
: Mike
: Greg Chaplin ([log in to unmask]) wrote:
: : Greetings.
: : We are implementing a system using Java to communicate between our HP3000 969 and an NT server to send data back & forth. I downloaded the latest version of Java in December 2000, and it was running happily in testing (well, other than 1 show-stopper, but that's another story). We then put up a firewall and locked down the ports being used, with the result that the 2 Java agents on the 2 boxes could not talk to each other, which resulted in the software being rewritten.
: : Now, I know virtually nothing about Java, so I'm no real help. The software now uses "rmid", an RMI daemon to register services being used (??). However, when I run rmid, it dies with the following:
: : shell/iX : rmid
: : Activation.main: an exception occurred: exception while taking snapshot; nested exception is:
: : java.io.IOException: System call error
: : sun.rmi.log.LogException: exception while taking snapshot; nested exception is:
: : java.io.IOException: System call error
: : at sun.rmi.log.ReliableLog.<init>(ReliableLog.java:116)
: : at sun.rmi.log.ReliableLog.<init>(ReliableLog.java:135)
: : at sun.rmi.server.Activation.main(Compiled Code)
: : The developer knows Java but nothing of the HP3000, so he's of little use.
: : I know nothing of Java, so I'm of no use.
: : Can anyone out there please shed a little light on this? Things to check? Switches to push or pull???
: : TIA,
: : Greg.
: <snip>
--
-----------------------------------------------------------------
Mike Yawn
Hewlett-Packard email [log in to unmask]
Commercial Systems Division Phone (916) 748-3865
8000 Foothills Blvd MS 5219
Roseville, CA 95747
-----------------------------------------------------------------
|
