| Author |
Thread Statistics | Show CCP posts - 10 post(s) |
|
CCP Habakuk
C C P
C C P Alliance
1062
   |
Posted - 2014.10.31 12:18:27 -
[1] - Quote
Progress is slow, but there is progress. From my testing in the last hour:
Good: Jumping is working again in the BrainInABox sandbox, even if you have no implants fitted (somehow it worked fine before with implants fitted).
Bad: The CPU of a ship is not modified correctly by skills on the client after boarding a ship in space (but it is modified on the server).
CCP Habakuk | EVE Quality Assurance | Team Gridlock
Bug reporting | Mass Testing
|
|
|
CCP Habakuk
C C P
C C P Alliance
1062
|
Posted - 2014.10.31 12:24:11 -
[2] - Quote
CCP Prism X wrote:Makari Aeron wrote:CCP Prism X wrote:It's our hope that moving things to C++ is not strictly necessary for this project. That doesn't mean it wont be it's own project. We'd just really really really want to release this in bits, harden them, and then do something more crazy. Releasing seven crazy-projects onto TQ at the same time is.. crazy. You devs are just a big box of crazy, amirite?  I'm just a programmer. I think the real insanity lies within CCP Habakuk who has been with Gridlock since it's original inception, and actually has to deal with the quality assurance side of our work. I can promise you that when you're polishing poop-code, you're going to end up with something of poop-ish quality. 
And CCP PrismX is posting about me without reading first the comment, which I just posted. 
CCP Habakuk | EVE Quality Assurance | Team Gridlock
Bug reporting | Mass Testing
|
|
|
CCP Habakuk
C C P
C C P Alliance
1067
|
Posted - 2014.11.01 01:40:33 -
[3] - Quote
Andy Koraka wrote:So serious question. We're all aware of how whenever 100+ pilots undock or jump through a gate, or otherwise undergo a session change the node gets bogged down and enters deep tidi (hence the need for BIAB in the first place). Will you guys be adjusting the node mapping to compensate?
A while back CCP changed the node mapping so that local space is generally on the same node, thus a fight in Delve wouldn't cause ti-di in Tenal or elsewhere spread across EvE.
Pre-Pheobe it wasn't really a big deal since most fleet travel was direct to the destination via jump drive/titan bridge, but Post-Pheobe fleets are expected to rely on taking gates the current mapping scheme is problematic.
...
I don't think that we will need to do (significant) changes to node mapping due to BIAB - but the load in many systems should be lower (we could increase the number of systems per node - but I don't think that we will do this). With BIAB it might be better to jump between systems on the same node compared to systems on different nodes - but it is a bit too early to tell as I don't have any numbers yet for this. Several years in the past we thought that it would be beneficial for server load to have neighboring systems on the same node - but mass tests (and reports from TQ) proofed this to be wrong.
For the jump changes with Phoebe: I can imagine, that we might see a few more small lag spikes for moving capitals, but time will tell how capitals are moving exactly after these changes. I don't think that it will be necessary to change the node mapping for this - but we will for sure monitor how the server is doing and take action if really necessary.
CCP Habakuk | EVE Quality Assurance | Team Gridlock
Bug reporting | Mass Testing
|
|
|
CCP Habakuk
C C P
C C P Alliance
1068
|
Posted - 2014.11.01 01:57:57 -
[4] - Quote
Sales Alt negrodamus wrote:PrismX, very nice work. I do similar work so I understand and appreciate the lunacy :D
Now at the risk of invoking the wrath of the parallelization gods, I gots a question that is probably ideally suited to you:
Given how the BIAB stuff is going to take a load of computation off a given node and put it elsewhere, would it be possible / technically feasible to setup an environment where a given solar system is effectively hosted on multiple nodes?
Even better, can we get dynamic node remapping?
You talking about a C rewrite makes me hopeful.
I'm trying to answer, as I am already active here in the forums:
As far as I can tell as non-programmer BIAB will not make it easier to host a given solar system on multiple nodes. We are trying to move as much load as possible from the system nodes to other nodes to reduce the need for splitting systems themselves (which would be a massive undertaking).
Dynamic node remapping: I would love it, but there is still a long way to go. BIAB might even help a tiny bit with this, but there is still a LOT to be done (and many other things are more important).
CCP Habakuk | EVE Quality Assurance | Team Gridlock
Bug reporting | Mass Testing
|
|
| |
|