I have used OpenFire at three client sites for almost four years now, and have never (not once!) had a problem! So please be gentle - my knowledge of OpenFire + Spark is limited, as it’s always “just worked” before.
I have a client site that I want to upgrade to Spark 2.6.0 when its released. I wanted to install Spark RC2 on a couple of desktops to see how users like it, but before I I did that, I installed it on a virtual machine. When I tried to log in, I got an “Invalid username or password” error. This is on a 32-bit XP virtual machine. It did not (nor never had) any version of Spark installed before today.
For testing purposes, I uninstalled RC2 and installed 2.5.8 and it works as expected. I then tried installing 2.6.0 RC2 over 2.5.8, but it loses the previous settings and gives the same username\pass error message. I don’t see anything of note in the server logs, but if I set up debugging in RC2, I get this:
I’m connecting via FQDN. I tried connecting by local machine name and by IP address and still get the “Invalid password” error.
Incidentally, I installed OpenFire 3.7.0 and Spark 2.6.0 RC2 on a VM here at the house last night for testing purposes, and the client connected without problem. I therefore thought the problem might be with the OpenFire install on the production server, so I exported the users to an XML file, uninstalled OpenFire (including deleting the leftover files\folders in %PROGRAMFILES%), reinstalled 3.7.0 and imported the users. Spark 2.5.8 works fine internally, as does Digsby and Imo externally. But Spark 2.6.0 still gives an “Invalid username or password” error.
Well. I’m hearing about such problem with newest version of Spark time to time here in the forums, but i have never seen this for myself both at my test and production servers.
Users could connect to the server initially but got the invalid username or password pop-up windows a few seconds later and could not log in even though the username and password are proven to be correct.
I had the exact problem. It would work under 2.5.8 but never after 2.6 or 2.6.3.
I managed to track down and solve the problem.
For some reason when I delete in the system properties or in the table ofProperty the xmpp.fqdn = server.domain.com setting, the new version of spark works correctly.
Thanks for the tip, John! I deleted xmpp.fqdn and the local 2.6.3 clients were able to connect… finally!
But after I deleted xmpp.fqdn, Digsby stopped working (which I installed on a few remote users’ machines). It’s no big deal to switch them over to Pidgin, but I wonder why Digsby apparently requires the FQDN property when Spark and Pidgin do not? (I’m not really expecting an answer here, more just “thinking out loud”.)
I also would like to know deleting the xmpp.fqdn or just entering the domain name (company.com) without the server address, what effects would it have.
So fa for me, it has not affected any operation of openfore, plugins or the clients.
We are usng Spark, ichat, and iphone clients without any issues.
Basically there is one bug in Openfire (whole description is on the forum and you can look on SPARK-751 ticket) and another BUG in SMACK (XMPP client that Spark uses to create xmpp connections )
Spark 2.5.8 is working because is using SASL PLAIN authentication, but Spark 2.6.3 is using SASL DIGEST
In Openfire - the SASL server is created based on the xmpp.fqdn value, but what is sent as digest-uri to the client is xmpp.domain
In Smack - the REALM is not correctly sent back to the client
I have patches for both problems attached on this forum and on SPARK- 751 as well
Sure thing - but I don’t have rights to open tickets/attach patches against Openfire or Smack (for smack there is an already opened ticket - SMACK-344 but I cannot attach the REALM patch to it)
I know Walter created experimental smack jars that include various fixes for next Spark release, which will be this month, it should include the REALM patch as well…
As you have mentioned above…I just want to know where could i find the **xmpp.fqdn **so i can delete it. as i have problem in login - in spark connection.