Originally by: Meissa Anunthiel

---

Originally by: Lord TGR

---

Originally by: Meissa Anunthiel

---

If a node supporting 10 solar system is at 95% capacity, everything will look fine and dandy for all those systems.

Your extra 40 people are most likely the proverbial straw that breaks the camel's back.

It has nothing to do with servers not being able to sustain fights of 40 people, it has everything to do with you being unlucky with the load distribution and the node not being reinforced.

---

That logic holds up if what we're looking at is memory or disk resources, which are very much finite, i.e. the difference between having spare resources or not is a harsh line. Usually involving crashing etc.

When it comes to CPU, however, that line becomes much more blurred, and I strongly doubt you'll go from "yay no problem changing system" to "oh my god I can't load system halp halp" just because you use slightly more CPU on that node. To go to halp halp-mode, you would have to use vastly more CPU during the migration from one node/system to another, and if that's the case then we're not looking at the proverbial straw, but the proverbial ton of bricks, and that changed radically in Dominion (and sounds like it got worse in Tyrannis).

However, I'm more inclined to think it's some sort of timing or threading problem, but it's damn hard to do anything other than random guessing as long as CCP is keeping quiet about it, which makes the whole deal vastly more annoying than if they had been forthcoming with information.

---

Agreed. It's more complex than just "CPU", it's resources in general (though they're CPU bound in this instance), though things tend to degrade very fast for everyone when timing/locking mechanisms can't happen in reasonable amount of time. While in Iceland, the CSM has had the opportunity of getting more information from CCP as to the cause of the lag but without knowledge of the system architecture, I doubt we can be of any assistance to find an architectural reason to the whole thing. However we can assist with the symptoms, ie, lag occurs when we do this, it occurs this way, I get this message/freeze/etc.

While it's nearly impossible for them to give us the information we'd need to help in solving the problem from that end, we encouraged CCP to provide us with information as to the challenges they face in trying to find a solution.

All I can say is that it's one tough problem that they're working on, and the people trying to solve it are not idiots...

---