113 replies to this topic

[Matthew Craig]

Post patch we had to react to failures in the Match-Making process. Following the patch, we saw an immediate change in behavior from both public and internal testing with hangs in the matchmaking logic.

After a couple of reboots to the Matchmaker, which players will have experienced as failed attempts to match, the issue continued to resurface. As such, our next step was to temporarily disable 3/3/3/3 to determine if this makes any change to the behavior. This appears to have stabilized the Matchmaker therefore, this means matchmaking weight class restrictions will not function until further notice.

We are actively investigating the cause and will update as we know more.

Thanks

[AlmightyAeng]

So, 3/3/3/3 broke the system. People started freaking out.

And you respond by quickly turning it off to find out what went wrong.

Good job! Thanks for the update.

Edited by Ghost Badger, 29 April 2014 - 12:32 PM.

[Dawnstealer]

What would likely correct the errors would be an Urbanmech. Urbanmechs fix everything. It's true: it's on the internet.

[DEMAX51]

Hope to see a hot-fix go up soon. From what I've heard from those who've been able to play in 'em, the 3/3/3/3 matches have been much closer.

[JohnnyWayne]

I'm neither suprised nor disappointed. I expected it because I understand how things are. But its funny to make stupid comments.^^

[AlmightyAeng]

DEMAX51, on 29 April 2014 - 12:32 PM, said:

Hope to see a hot-fix go up soon. From what I've heard from those who've been able to play in 'em, the 3/3/3/3 matches have been much closer.

Ran into quite a few rolls, including a 12-0 on the test server...but I'm hoping that was just ELO being unable to cope with very tiny population buckets.

[Dawnstealer]

Yeah, they'll find and fix it. I'm actually looking forward to 3/3/3/3, so once it's up and running: cool. Glad they caught it quick - I'm not even off work yet.

[Redshift2k5]

3/3/3/3 was nice on Public Test.

I really wonder why the patch is acting differently on the real server vs the public test server we just played a few days ago. Is it the game build or is it just player metrics? We all failed to find matches for a while on PTS until it got on more steam.

[Jman5]

Redshift2k5, on 29 April 2014 - 12:37 PM, said:

3/3/3/3 was nice on Public Test.

I really wonder why the patch is acting differently on the real server vs the public test server we just played a few days ago. Is it the game build or is it just player metrics? We all failed to find matches for a while on PTS until it got on more steam.

I don't believe it is acting any differently. On the public test server there was an infinite search time so fail to find never happened. However many people, myself included, were still getting caught in infinite search time loops until they canceled and restarted the client. Despite the fact that I have helped out with several of their test clients without issue, they insisted that this time it was low-population that was messing with search times. Clearly they were wrong.

Edited by Jman5, 29 April 2014 - 12:45 PM.

[MischiefSC]

It was awesome. Wants it back, please. Soon as you can.

Until then...

It's still all so PRETTY with no film grain/glass/DOF!

[Bilbo]

Jman5, on 29 April 2014 - 12:43 PM, said:

I don't believe it is acting any differently. On the public test server there was an infinite search time so fail to find never happened. However many people, myself included, were still getting caught in infinite search time loops until they canceled and restarted the client. Despite the fact that I have helped out with several of their test clients without issue, they insisted that this time it was low-population that was messing with search times. Clearly they were wrong.

Bilbo, on 29 April 2014 - 12:44 PM, said:

I played on the PTS from 6pm EST until 1am EST. Never once failed to find a match, wait time was quite long at the end though.

[Gyrok]

Err...yes...?

...

Tonnage limits instead?

Edited by Gyrok, 29 April 2014 - 12:47 PM.

[Matthew Craig]

Jman5, on 29 April 2014 - 12:43 PM, said:

I don't believe it is acting any differently. On the public test server there was an infinite search time so fail to find never happened. However many people, myself included, were still getting caught in infinite search time loops until they canceled and restarted the client. Despite the fact that I have helped out with several of their test clients without issue, they insisted that this time it was low-population that was messing with search times. Clearly they were wrong.

This is definitely different Jman5 we're not sure why exactly ourselves, the match making process didn't experience any hangs on Public Test but was hanging until 3/3/3/3 was disabled on Live we're working to determine why.

[Pyrrho]

Matthew Craig, on 29 April 2014 - 12:48 PM, said:

This is definitely different Jman5 we're not sure why exactly ourselves, the match making process didn't experience any hangs on Public Test but was hanging until 3/3/3/3 was disabled on Live we're working to determine why.

There is a missing semi-colon somewhere in the code!

[Mofwangana Bogogono]

Dawnstealer, on 29 April 2014 - 12:32 PM, said:

What would likely correct the errors would be an Urbanmech. Urbanmechs fix everything. It's true: it's on the internet.

Dawnstealer should be on the development team. The man is a genius.

[MischiefSC]

Matthew Craig, on 29 April 2014 - 12:48 PM, said:

This is definitely different Jman5 we're not sure why exactly ourselves, the match making process didn't experience any hangs on Public Test but was hanging until 3/3/3/3 was disabled on Live we're working to determine why.

So I got into 1 match just fine right after patch. Both teams full.

Then I got 'failed to find match'. I attempted to exit but when I went back to the login screen (atlas, login/password) it said 'connecting'.

So I closed the client and restarted. It showed my mech, which had been getting 'failed to find a match', said it was in a match? Then I dropped, got a match and most the other team was discoed.

It's like I was getting 'failed to find a match' but in fact my mech was dropping without me. Clearly excited to enjoy the 3/3/3/3 goodness and the lack of film grain it started without me.

Hope you find the fix soon! 3/3/3/3 was awesome.

[Helmer]

Jman5, on 29 April 2014 - 12:43 PM, said:

Despite the fact that I have helped out with several of their test clients without issue, they insisted that this time it was low-population that was messing with search times. Clearly they were wrong.

Between people playing with private matches, and only a fraction of the population patched up yet, I don't think we have the data to say they were clearly wrong. Unless you're privy to some data the rest of us are not?



Cheers.

[Heffay]

Pyrrho, on 29 April 2014 - 12:52 PM, said:

There is a missing semi-colon somewhere in the code!

Obviously. I can tell that from way over here!

[Jman5]

Matthew Craig, on 29 April 2014 - 12:48 PM, said:

This is definitely different Jman5 we're not sure why exactly ourselves, the match making process didn't experience any hangs on Public Test but was hanging until 3/3/3/3 was disabled on Live we're working to determine why.

Alright, well I just hope you guys get to the bottom of it so we can try out the new 3/3/3/3 teams.

[Jman5]

Nikolai Lubkiewicz, on 29 April 2014 - 12:56 PM, said:

On a side note: We are experiencing an issue in this patch in which players who had selected ANY were not being matched with those who selected a specific game mode. This resulted in increased wait times for both sets of players. Our quick fix to this issue for the patch was to hard lock the public match queue to ANY. For now, players joining public matches will be unable to select Game Modes.

We are expecting a hot-fix for this issue ASAP, along with corrections to the logic and re-implementation of 3/3/3/3

Is it at all possible that people who had previously deselected one or two game modes prior to the patch were still hard locked into that setting?