Started reviewing the upgrade to 3.10.2 of Openfire today and found that it is recommended to upgrade clients to the new spark 2.7.1 prior to server upgrade otherwise clients may not be able to communicate with the server. To accomplish this, I am trying to load the client version into openfire for deployment to clients.
But when I test logging into spark on a system with spark 2.6.3 it is not prompting for the available update. Additionally, when I manually install 2.7.1 and logon, it is not showing up as 2.7.1 under my sessions tab instead still listing as 2.6.3.
Is there something that I am missing?
Does Openfire 3.9.3 recognize the 2.7.1 release info when deploying to clients?
Will I have to push the client update a different way than the openfire client version check?
I have approx 250 users for spark so I would prefer limiting the amount of extra work on my part for deployment.
Spark 2.6.3 should be able to login to Openfire 3.10.2 (UPDATE 2: Openfire 3.10.0 Released). Though i still advise to stay up to date and upgrade to the latest Spark version.
If you are trying the regular “Check for updates” menu, which should by default check for new versions on igniterealtime.org, then this won’t work as the updating code is very old (legacy code from JiveSoftware, when Spark was their proprietary product) and points to not existing web resources. I have filed a ticket, but this probably won’t be fixed (most probably updating option will be removed). https://igniterealtime.org/issues/browse/SPARK-1623
This option still works though if you are using Client Control plugin on the server. But it only works for the same major versions (2.6.x, or 2.7.x). It can’t update 2.6.3 to any 2.7.x versions as the version scheme has changed and the updating code hasn’t been updated as i said. I was able to push 2.7.1 version to 2.7.0 client this way. So it can be used in future. Though, a user has to have administrator rights to be able to install new version, which is highly improbable in the enterprise environment (one more point to remove this feature as it is mostly useless). One more thing to remember, an executable name can’t have spaces in it, when uploading to Control Client page as Spark won’t be able to run the installer.
Version shown on the Sessions page is a misunderstanding. It is not a version of a client, it is a resource handle of a logged in user. Spark is putting in its version as a resource by default for the first login of a user, but it doesn’t update it when version changes. I have removed adding version number into a resource in 2.7.1 version (now it defaults to just “Spark”), but all existing users would still have “Spark 2.6.3”, unless you manually change it for everyone. 2.7.2 will have an option to use the current version as a resource (which will update when version changes), so it would behave as you expected. https://igniterealtime.org/issues/browse/SPARK-1619 But you would have to enable this option for every user.
Now, for the deployment, if you have AD, you can use GPO scripts to do this. I run a batch file on shutdown to update Spark silently. You can find more information here Trying to use GPO to install Spark