| Author |
Thread Statistics | Show CCP posts - 7 post(s) |
|
CCP Masterplan
C C P
C C P Alliance
1618
   |
Posted - 2014.09.04 13:43:00 -
[1] - Quote
We've seen in increase in reports of client 'black screen' lock-ups recently, particularly from players running multiple clients. Typically the black screen seems to occur when multiple clients are logged in from the same PC at the same time, and may coincide with mass dock/undock/jump session changes.
I'm investigating this issue and would like your help if you are experiencing this. Please note that other crashes or graphical lockups are also important to investigate, but I'm specifically looking for focused feedback related to this particular issue only in this thread: Multiboxing users that suffer black screens on some of their clients when those clients all attempt to login/jump/dock/undock at the same time.
If your clients suffer from this problem (especially if you've noticed an increase recently) please reply to this thread with the following info. I'm chasing up a few leads and this would help me to a) isolate the reproduction conditions and b) create a solution.
1] How many clients do you typically use
2] How frequently does the issue happen on Tranquility?
3] Are the characters in player corps or NPC corps?
4] Are they logging in to space or to a station?
5] Can you reproduce the issue with the same setup on Singularity
I'm particularly interested in anyone who is able to help reproduce this on Singularity, as there's things we can experiment with on there that we obviously couldn't do on TQ
Thanks for your help!
"This one time, on patch day..."
CCP Masterplan -á| -áTeam Five-0: Rewriting the law |
|
|
CCP Masterplan
C C P
C C P Alliance
1618
|
Posted - 2014.09.04 13:55:00 -
[2] - Quote
Logins that are simply slow but eventually get there are most likely unrelated. These black screens I'm investigating seem to be unrecoverable without restarting the client
"This one time, on patch day..."
CCP Masterplan -á| -áTeam Five-0: Rewriting the law |
|
|
CCP Masterplan
C C P
C C P Alliance
1619
|
Posted - 2014.09.04 15:52:00 -
[3] - Quote
Thanks for your prompt feedback. The fleet-finder availability (or lack of) is quite useful info here. I've also seen someone else mention that the agent tab in the station info window also fails to display information if a client is broken in this way. This is making my spider-sense tingle :)
RoCkEt X wrote:1] Ergo, this issue is something to do with being in space, and not to do with session change (because you can still switch ships whilst docked - thus session changing without an issue)
Just a quick comment on this: Switching ships whilst docked is no longer a session change (It used to be, until a couple of years ago). This is why you can switch quickly between different ships in your station hangar without requiring a 10s session timer between each.
"This one time, on patch day..."
CCP Masterplan -á| -áTeam Five-0: Rewriting the law |
|
|
CCP Masterplan
C C P
C C P Alliance
1619
|
Posted - 2014.09.05 11:34:00 -
[4] - Quote
Thanks everyone for your reports so far. I've made some changes to the build we have just updated to Singularity. If anyone who was previously able to reproduce the issue there is able to update their test clients and see how it behaves now, I'd be very interested in your feedback.
"This one time, on patch day..."
CCP Masterplan -á| -áTeam Five-0: Rewriting the law |
|
|
CCP Masterplan
C C P
C C P Alliance
1622
|
Posted - 2014.09.09 13:53:00 -
[5] - Quote
As listed in today's patch notes we've put out what should be a fix for this particular issue. I'd appreciate feedback from those affected if it helps your clients to behave.
"This one time, on patch day..."
CCP Masterplan -á| -áTeam Five-0: Rewriting the law |
|
|
CCP Masterplan
C C P
C C P Alliance
1622
|
Posted - 2014.09.09 15:55:00 -
[6] - Quote
It was caused by a concurrency issue when multiple clients made requests for certain server resources simultaneously and tried to cache the results, that only triggered under very particular cache states. The caching operations of one client could then corrupt the state of another, if that second client tried to read from the cache at a critical moment.
Ironically I think that the more you worked to reduce system latency, the higher the chance of encountering this bug might have been. A slower PC that had to hit the disk more often was probably less likely to trigger it, as there was less chance of two clients performing the necessary sequence of operations concurrently.
"This one time, on patch day..."
CCP Masterplan -á| -áTeam Five-0: Rewriting the law |
|
|
CCP Masterplan
C C P
C C P Alliance
1624
|
Posted - 2014.09.10 11:18:00 -
[7] - Quote
Rain6637 wrote:by the way, I left log server running last night, and ended up with 6.3 gigs of errors.LBW files (323 files). no blackscreen events, though. just normal operation. do you want them?
I had the client drops at 05:05, and i'm submitting a bug report with my log server files. other than that, I went through several session changes with everything running with no blackscreens. if you'd like those logs anyway, i've saved them... lemme know o/
The disconnects at 5am were due to unrelated network issues with our datacenter. As long as things are otherwise working for you, I'm happy to not have to go through several gigs of logs :)
"This one time, on patch day..."
CCP Masterplan -á| -áTeam Five-0: Rewriting the law |
|
| |
|