hm i did a test with 4.6.6. When i manually triggered the reconnect via admin console and switched to the sessions tab i could see that clients which where connected to 5222 got the detached state (stream management active). What about 4.7.? Same behavior?
With 4.7 launched manually or automatically the processes are still appearing as invalid in session tab and the number or Invalid sessions are increasing.
The behaviour using your plugin seems diferent as if I force the “same” process using openfire in the original way.
My only option is to force it manually. Once done in this way the Invalid conection starts disappearing
ok i think just stopping and reenabling the Connection Listener is not enough in 4.7.
Strange…
The only thing i can do is to reinit the listener with a configuration…
i have moved some outputs to debuglog … you could enable the extra logging switch and enable debug logging in openfire to see all logs…
otherwhise it now only logs if restarting the listener…
1.1.3 now sets the config values again… hope it will help
But should still be used as an workaround…
Could you provide more details:
OS, Java Version, use of StartTsl / legacy ssl / …, ssl cert is wildcard / self signed …?
I think as much Infos as possible is needed to
track down the problem
Maybe some Screenshots of the configpage for client connections… do you use all krypt algos…?
We seem to have identified one server that is having this issue, but which is also suffering from a corrupted HSQL (embedded) database. I wonder if those problems are related.
I’m not sure how to properly check for database corruption. What we found was that, using the “DB Access” plugin, executing this query would result in an (almost) blank screen:
select payload from ofPubsubItem;
The logs contained an error that included the wording: “lob is no longer valid”
This leads me to believe that the HSQLDB file that stores LOBs is no longer ‘in sync’ with the rest of the database files for HSQLDB.
Can someone using the embedded database check for the same issue?
i am using sql server and have theese issues with invalid sessions too … using OF 4.5.x branch so i think it is not absolutely related to the embedded database
hi Guus, it’s me, vladimir, from conference. i told you about this problem earlier. in 4.6.3 i have not this problem, but in one time at nightless before 4.6.4 it happened. closer to 4.6.4 release.
i use ubuntu, mysql, freessl.
I have gone over all changes between 4.6.3 and 4.6.4, but nothing stands out as a likely cause for this. I’m somewhat at a loss. Some long shots:
Can someone please check if setting the system property xmpp.client.version-query.enabled to false will prevent the problem?
Do people, long before the problem occurs, see this in their log files? Rejecting stanza from client that has not (yet?) established an authenticated session