Unusual Lag Spikes?

[Mature]

I noticed while crafting that every time I lag is usually around 0000 hours (midnight) UK time.

This has been happening for over few days. Hope they look into it

[GTANJ]

I haven't (I believe) experienced any serious lag yet since I started playing. Can anybody describe what the lag looks/feels like?

[bounddreamer]

Half my raid group is still experiencing lag, rubber banding, and DCs. We actually had to stop practice Tuesday night because we couldn't deal with resin placement during the jump. We were totally shocked to actually clear last night because our first few attempts were plagued with lag.

[Dotsusama]

Still getting lag spikes, but I'm not with Comcast (Shaw is my ISP, western Canada). Nothing to report from their end, and I don't experience lag any where else.

It would be like this video here. At about 20 seconds you can see everyone stop, and then speed up to get to where they would be if they had kept running. Make that happen every 4-5 seconds and you have my experience in PVP and Trials all this weekend; to say nothing of trying to get around the city or open world.

Another symptom is having instant buffs show up 1-2 seconds after hitting the buttons. I've NEVER had this before. Example:
>Hit Swiftcast (Controller)
>Mouse over and click Summon Pet Macro
>Summon Pet hardcasts slow
>2 seconds later, Swiftcast starts ticking down on the buff bar

[Dotsusama]

I knew I forgot to include something, so here's my fresh tracert from tonight (after step 2). It looks the same even if I use the IP from before, 199.91.189.35

Tracing route to 199.91.189.45 over a maximum of 30 hops

  3    10 ms    11 ms     9 ms  te-1-12-0-12-ar01.area4.il.chicago.comcast.net [
68.87.232.93]
  4     *        *        *     Request timed out.
  5    30 ms    30 ms    31 ms  ae-10-10.car2.Montreal2.Level3.net [4.69.153.86]

  6    30 ms    31 ms    30 ms  ae-10-10.car2.Montreal2.Level3.net [4.69.153.86]

  7    30 ms    31 ms    30 ms  ORMUCO-COMM.car2.Montreal2.Level3.net [4.59.178.
74]
  8    28 ms    28 ms    27 ms  192.34.76.10
  9    28 ms    41 ms    28 ms  199.91.189.242
 10    29 ms    27 ms    28 ms  199.91.189.45

Trace complete.

[Raist]

I knew I forgot to include something, so here's my fresh tracert from tonight (after step 2). It looks the same even if I use the IP from before, 199.91.189.35

Tracing route to 199.91.189.45 over a maximum of 30 hops

  3    10 ms    11 ms     9 ms  te-1-12-0-12-ar01.area4.il.chicago.comcast.net [
68.87.232.93]
  4     *        *        *     Request timed out.
  5    30 ms    30 ms    31 ms  ae-10-10.car2.Montreal2.Level3.net [4.69.153.86]

  6    30 ms    31 ms    30 ms  ae-10-10.car2.Montreal2.Level3.net [4.69.153.86]

  7    30 ms    31 ms    30 ms  ORMUCO-COMM.car2.Montreal2.Level3.net [4.59.178.
74]
  8    28 ms    28 ms    27 ms  192.34.76.10
  9    28 ms    41 ms    28 ms  199.91.189.242
 10    29 ms    27 ms    28 ms  199.91.189.45

Trace complete.

With that redacted report, would have to say someone needs to investigate what is going on at the exchange point into Level3. Typically the reason it is enforcing a rule to not respond to ICMP at such a hop is because of imminent congestive failure. If you ever see a response from hop 4 when you run a trace, this is most likely the case.

If there is congestion or bad jitter at the omitted hops 1 or 2 it can also be impacting things further down. It is all cumulative. So if there is bad congestion locally and then again at somewhere like Vancouver/Ontario/Toronto along a Canadian path, the Great Lakes/Ohio/North East for northern US path, or Southern Cal/Texas/Kentucky otherwise (not sure exactly how far west you are, common points of congestion we've seen), it can be compounding the issue.

Also, keep in mind one of the drawbacks to low latency: because it scales against your normal latency, it doesn't take nearly as much variance to lead to an out-of-order packet. That can trigger retransmits and lead to stuttering/rubber-banding as well, even though your latency is normally low. In that sample we are seeing right at a 50% variance being triggered by something in groups of only 3 pings. Not a very good sign.

[Dotsusama]

snip

I just cut the first two out of habit for public posting, both those first two hops are consistently 8~10ms each. I think out of the dozen or so times I've run this I've seen a response and an actual IP from 4 like... once. I even did a search on it so thankfully it was still in my history! It's another Level3 at 4.68.63.125.

EDIT (I keep doing this...): I mentioned it before, but I'm in Chicago proper, like some other users here with the complaints, and using Xfinity/Comcast. I did the tracert at work which is just a couple of blocks away, and that office uses AT&T Uverse DSL and STILL has the same damage at step 4.

[Raist]

yeah... that's a troubled spot for them too. Long/Lat (38/-97) in the geo data puts it around Wichita. Seen them report problems there frequently. Would hazard a guess that may be a solid point to investigate.

Looks like Level3 is still dealing with some downed circuits in general. Reports of one going down in Dallas today, and a sort of "rolling blackout" effect seems to be happening in California today as well. There was a tweet that they were working with TWC to address one circuit about 8 hours ago. Doesn't show where though. Could be along that Louisville/Cincinnati/Columbus/Cleveland run that has been flaking out for a little over a week now.

Level3 just seems to be knackered up in general these days...more so than usual. IDK why some ISP's are so reluctant to switch away from them, even if just for 30 days or until their next announcement cycle kicks off.

[MihaelB]

Keep DCing randomly, not my computer's problem or internet's problem either....

[Raist]

Keep DCing randomly, not my computer's problem or internet's problem either....

Did you check every hop between you and SE's servers? Hard to say with such certainty that it isn't an internet related problem without looking closely at the route.

Edit:
Just took a quick look at my route. Signs of heavy congestion here in South Carolina (around the Conway/Myrtle Beach area--high tourism/retail area), and again showing up around the New York/New Jersey segments. Not enough to trip me up much yet, but the potential is there if packet loss starts happening.