[jcifs] Weblogic 6.1 and jCifs failure on redeploy
eric.glass at gmail.com
Thu Aug 19 10:05:27 GMT 2004
> What can be happened that the SMBSession couldn't be found after the redeploy
> of the application?
Not sure; does it work if you restart the WebLogic instance?
Redeployment may have a variety of strange effects on jCIFS. If I had
to guess, I would say that redeploying the app is probably
"reinitializing" jCIFS but maintaining the user's session. If this is
the case, then something like the following is likely happending:
1) User logs on, and the NtlmPasswordAuthentication object is stored
in the session.
2) Application is redeployed; jCIFS is reinitialized, and the
connection to the SMB server is severed (or at least jCIFS no longer
knows it exists).
3) Subsequent HTTP request from the user contains the session cookie,
which still maps to a valid session (possibly serialized/deserialized
during redeployment of the app).
4) The (now invalid) NPA object is retrieved from the session, and
jCIFS passes the request on without reauthentication.
5) jcifs-ext "flips out" with the invalid NPA.
> The second problem is more jcifs-ext specific. When I'm requesting the groups
> of the authorised user and one group has more than 20 characters the request
> get failed with result code 234 (more data available). Is it possible to get
> groups with more than 20 characters length?
This requires RPC (which jCIFS/jcifs-ext doesn't support yet).
jcifs-ext uses RAP calls, and the relevant field in the RAP version of
the GROUP_USERS_INFO_0 structure can only hold 20 characters (actually
21, but the last one is a null-terminator).
More information about the jcifs