we are currently running a Jboss 4.2.3 on Linux (Suse SLED 10) with an Oracle 10g Database and Java6 (1.6.0_11-b03).
We need to create a folder structure that a customer should have acces to.. Folder names are coming from a oracle table.
Unfortunately there is an encoding problem when creating the folders on the filesystem, all umlauts ('ÃƒÂ„ÃƒÂ–ÃƒÂœÃƒÂŸ') aren't correctly converted but instead are shown as question marks ('?'). The linux machine is using reiserfs as filesystem and in the last days I have used every combination for the local encoding in ".profile" (Bash) and in "run.conf" (JAVA_OPTS): de_DE.iSO8859-1, de_DE@euro, en_US.ISO8859-15
What I have seen is that the jvm start parameter for file.encoding is not working with JBoss. I can see in boot.log that the parameter for file.encoding is logged but in the running jvm the file.encoding is always set to "ISO8859_1" (not ISO-8859-1). In a jvm that is run like "java Test.testclass" the file.encoding is correctly set according to settings in .profile or start parameter.
The same problem happens in jboss' own logfile, all umlauts are logged as question marks too.
What we can say is though, that the database isn't the problem as its no problem to select lines with parameters that are containing umlauts or write (update) lines with umlauts. Also, the byte array of an string that contains results from database shows that the correct byte is used for umlauts.
Currently we are running out of ideas what's the problem here. Can someone help with a hint please?