[FS |HSO|RaideR]

Actually they are both right, if Urban Terror pulled enough funding to be hosted in such an environment, we would be there.

For the record, our host is OVH, and we now have ESXi VM Images of all our network. We can start them in the OVH Cloud if we need too or simply push them else where such as other hosts.

Urban Terror is both Bandwidth and Database Heavy. Both of which in cloud based solutions end up being outside our budget limits. There are better ways of hosting Urban Terror and better ways of delivering the services. But this is the best way we can afford based on the advertising the website provides.

I don't mind the critique at all, but we work with what we have. The "golden" network for urbanterror is at the moment "outside our means".

[e-junkie]

thanks for your continued hard work over the years
i'm starting to get to an age where you can afford to spend a few bucks here and there so i want you to know that i think you can tell us if you need some financial support. i'd throw some in for sure.

[thelionroars]

ItsMe, on 30 November 2014 - 11:09 AM, said:

You're totally right, those guys at Amazon, Google, Dropbox (who host on AWS) are some nobodies with no education and grades compared to RaideR and EvilElf with no knowledge about redundancy, clusters and disaster recovery.
Hint: AWS and co. are build for exactly that. Those few CPU cycles the database/ master/ auth need are a cheap buy there.

I was making exactly the same point you went on to make that everyone has their budget limits. Take some of your own advice about not barking like a kicked dog

[Kenny]

Yeah, I wasn't trying to crap on Raider or Elf. From what I understand there more network/IT guys anywho...

I can sympathize with the pain they went through getting it up again, which is why I suggested some cloud providers. There may be some items urt can cheap out on to get on board a cloud provider. Some of the software services that are there are a current trend these days for a reason.

This post has been edited by Kenny: 01 December 2014 - 05:15 AM

[TehReaper]

Thanks for all the hard work guys! Hardware failures are never fun.

[Angelface]

thelionroars, on 29 November 2014 - 12:05 PM, said:

I'm sure RaideR (who makes a very respectable living doing admin)

Read: Helpdesk, and by the look of it, Tier 1.

thelionroars, on 29 November 2014 - 12:05 PM, said:

and EvilElf (who holds a CCIE)

Did he get laid off by Cisco with their data-center reconstruction, too?

I'm sure you're well-meaning, but between a "respectable" admin and a CCIE, the site just can't seem to stay up. It's enough to make one curious as to exactly why it went down in the first place.

I think both praise and criticism have a place here. Yet I see nothing but praise.

[|P|Nitro]

I am not going to say that Frozen Sands have the perfect infrastructure in place, because quite frankly they don't. However they also don't have the financial resources to develop such a infrastructure that has good high availability in place. Thatis not to say they do not work with what they have really well!

Sometimes people forget how much maintenance can be involved behind the scenes and especially with the continuous development of both UrT4.2 and UrT HD I can't say a single bad word about how the recovery of the website and services was tackled.

People forget that everyone on the team is working for free and also have full time jobs and family commitments that take priority first (even during disaster recoveries)

I think it is selfish for one person to compare the downtime and recovery of urban terror to that of a proper business model with the likes of battle.net (which even has its own problems too).

In fact the time taken to recover from hardware failure was actually commendable I thought given that a small volunteer group only working a few hours each night was able to have the website and other services back up and running after a few days.

Also lets face it, The website being down a week or two doesn't really affect the players in a major way. Game Servers were still up, people still played like normal it just required a little bit more know how than the usual pick a server off a list.

And lastly, targeting admins and developers for not working to your time scale is laughable at best but quite simply moronic, targeting their skills is another pointless discussion. They have built and maintained this game for over a decade - what have you done?

The only single point I would like to talk about though is to have TWO separate master servers on different physical machines/locations that way there is some extra redundancy for the players compared to the current situation were servers send double broadcasts to the same machine just different IP's

[Acid3croco]

Barbatos, on 30 November 2014 - 03:33 PM, said:

Please check the value of your cvar cl_master. It should be "master.urbanterror.info". Then click on "Get new list".

Hello,
And how do i do that since my console is all blurred in red, blue and white colours...

This post has been edited by Acid3croco: 02 December 2014 - 05:49 PM

[FS Barbatos]

Nitro, on 02 December 2014 - 02:34 PM, said:

The only single point I would like to talk about though is to have TWO separate master servers on different physical machines/locations that way there is some extra redundancy for the players compared to the current situation were servers send double broadcasts to the same machine just different IP's

We decided to launch two new official master servers that are on different machines (one being physically hosted by a different provider in a different place). master2.urbanterror.info is up and running and master3.urbanterror.info should be up in a few days.

We also modified the engine source code to fix two game issues with master servers:
- the client can only query one master server (cl_master), you have to manually change the cvar to try other masters - fixed here. It is now trying to query the master from the cvar cl_master first, then tries master/master2/master3.
- the game servers do a DNS lookup of the master server IPs only once at server launch, which implies that server admins have to restart their servers in case of a master server IP change - fixed here. The servers are now doing a DNS lookup of the master server IPs at each map load.

These fixes will be applied in 4.2.022 and I'm now fairly confident to say that we shouldn't have master server "general shutdowns" anymore.

[|P|Nitro]

now thats some great news right there! thanks for the heads up barbie :)