jabber.hot-chilli.netWe experienced problems with ICQ transport the last days. According to the developers of Spectrum this is a “glib poll” issue. epoll unfortunally does not work with glib (yet).
Therefore Spectrum ICQ transport now uses libev3 and epoll.
It looks good at the moment, I hope it stays that way. We will see what happens during the next days as our server is the only server which runs this configuration. Not even the Spectrum developers were able to test it, yet. ;-)
We just restarted the server to activate Remote roster. What is that?
What’s Remote roster?
Remote roster is protoXEP which allows your server to synchronize legacy network contact list with your users’ rosters. Currently it works only for local users (users registered on the same server which supports this protoXEP). It’s available for Ejabberd 2.1.x and Prosody.
Why should I use it?
It allows to your users following features:
- When they register their account and login for first time, they won’t receive lots of subscribe presences or Roster Item Exchanged stanza. All their legacy network buddies just appear in their roster with proper groups and nicknames.
- When they will unregistering their account, buddies are just removed from their rosters without need of unsubscribing them.
- When they rename some buddy in their Jabber roster, it’s renamed also in legacy network.
- When they move some buddy into different groups, it’s moved also in legacy network.
- All changes made in legacy network when transport is disconnected are synchronized with Jabber account once they become online in Jabber.
Taken from Spectrum website.
…happened the last days. This occured out of nothing. Same symptomatic like some time ago when the Jabber server thread suddenly took all available resources and killed the server itself.
We digged deep into the problem and hope, we found a solution for it – which we implemented about one hour ago. A short service outage took place because of a restart to activate the setting.
Hopefully we got to the problem.
Since today there is a new transport online: Identi.ca
The JID is identica.jabber.hot-chilli.net.
Have fun with it! :-)
Sorry for posting just now. Facebook transport was blocked for Facebook use only on the day we posted about it, which means October 6th. We did not delete the old contacts for local users, as the user should get unsubscribe messages as all other users. If you still have old users in your roster, please delete these entries manually.
The second XMPP transport is online since since October 6th as promised.
As mentioned here we finally have the feature which allows us to limit our Facebook transport for Facebook use only. We will delete all non Facebook accounts from the Facebook transport later today. You will be able to recreate these accounts on our second and new XMPP transport called xmpp.jabber.hot-chilli.net. We created this second XMPP transport because we noticed that more than one XMPP transport seems to be important for many people.
If you used Facebook transport for general XMPP usage and are a local hot-chilli Jabber user we delete all old contacts containing @facebook.jabber.hot-chilli.net in your roster for you. If you are a non-local user you might have to do this by yourself – which means that you have to delete all contacts containing @facebook.jabber.hot-chilli.net. When unregistering the accounts Spectrum sends unsubscribe notices to the Jabber servers, local and non-local. But all of you (local and non-local users) have to delete any leftover and obsolete contacts which might be left in your roster and/or the Jabber client.
We will announce when we finished the chances on the server here in the blog.
To activate the second XMPP transport the Jabber server will be restarted at 5:50 CEST today.
Are we under attack again? It seems – because the server went down yesterday and today. Sorry for the outage, we are looking into the logs as I write this post. Maybe we can find something…
Since our provider now supports native IPv6 (including reverse DNS) we changed our IPv6 configuration and now no longer use an IPv6 Tunnel Broker. Therefore we had two short service outages today to reboot the machine and activate the new settings.
If you are using IPv6 you might have to flush your DNS (or wait up to 24 hours from now) to get the correct IPs back from DNS.