Web sQuad ISP - Feedback Thread #2

Do you have any specific examples? Have you checked for packet loss? Transit interfaces are looking consistent this morning. And tests to known locations look good.
I have raised a ticket with support regarding this matter and have shared some test results

Refer to Ticket ID: #284505
 
Same here. Internet seems sloww on my CISP and Websquad accounts
Log a fault if you're struggling with specific services. A reminder that even though WACS is back, some peered services like Microsoft are not back to 100% yet.
 
Uhh @websquadza

1695485586742.png

Been a few days now, where I'm getting periods of ~15-30s of net just dying entirely a couple of times a day. Just had another one now
1695485702164.png

1695485731679.png
 
Had a PPPoE dropout at 6:30 to 6:34 in Randburg
 
Had a PPPoE dropout at 6:30 to 6:34 in Randburg
What is going on this morning?
Same here. And the F1 is about to start.

Morning we did notice this morning’s drop. We are investigating it. Currently monitoring closely.

Uhh @websquadza

View attachment 1591798

Been a few days now, where I'm getting periods of ~15-30s of net just dying entirely a couple of times a day. Just had another one now
View attachment 1591800

View attachment 1591802
I do see a small blip; we are trying to see if it’s related to today. More updates to follow
 
Morning we did notice this morning’s drop. We are investigating it. Currently monitoring closely.


I do see a small blip; we are trying to see if it’s related to today. More updates to follow
Another one now

1695550717755.png


1695550722667.png


1695550731255.png

1695550741832.png
 
We just saw one of our CEs have a bit of a traffic processing issue. We are investigating what the cause was. Team is closely monitoring to see what is causing this issue.

Edit: it looks like we are picking up floods of broadcast traffic on one of our vendor NNIs - this traffic is flooding the processing capability of some devices, causing these short interruptions. Our team has reached out to the vendor and is also working on a workaround to reduce/eliminate broadcast across this NNI. Please bear with us while we implement changes here - a large number of services need to be modified.

Edit 2: we have applied an interim solution as we've been able to further isolate the issue to specific VLANs. It looks like the vendor is not processing these vlans correctly and in some cases inadvertently bridging them, resulting in these broadcast loops.
 
Last edited:
We just saw one of our CEs have a bit of a traffic processing issue. We are investigating what the cause was. Team is closely monitoring to see what is causing this issue.

Edit: it looks like we are picking up floods of broadcast traffic on one of our vendor NNIs - this traffic is flooding the processing capability of some devices, causing these short interruptions. Our team has reached out to the vendor and is also working on a workaround to reduce/eliminate broadcast across this NNI. Please bear with us while we implement changes here - a large number of services need to be modified.

Edit 2: we have applied an interim solution as we've been able to further isolate the issue to specific VLANs. It looks like the vendor is not processing these vlans correctly and in some cases inadvertently bridging them, resulting in these broadcast loops.
The good old turn off spanning tree because it's rubbish? Then not have any alternative path protection. Which FNO???
I can have a guess.
 
The good old turn off spanning tree because it's rubbish? Then not have any alternative path protection. Which FNO???
I can have a guess.
Weirdly not this time. Arista has a very solid STP implementation (better than Cisco I’d argue) and we use strict MSTP configs to avoid broadcast leaks such as those seen between the Vuma and SADV or Frogfoot and CCC’s networks in the past. It wasn’t either of those FNOs this time. It’s DFA, first time seeing this kind of thing from them, in over 8 years of working with them. Broadcast wasn’t on the port or S-TAGs either. It was on the inner CTAGs. Seen something similar on another FNO (also on Ciena kit) in the past, but they controlled the broadcast they caused and it didn’t flood back, it just dropped traffic within their network.
 
Last edited:
Weirdly not this time. Arista has a very solid STP implementation (better than Cisco I’d argue) and we use strict MSTP configs to avoid broadcast leaks such as those seen between the Vuma and SADV or Frogfoot and CCC’s networks in the past. It wasn’t either of those FNOs this time. It’s DFA, first time seeing this kind of thing from them, in over 8 years of working with them. Broadcast wasn’t on the port or S-TAGs either. It was on the inner CTAGs. Seen something similar on another FNO (also on Ciena kit) in the past, but they controlled the broadcast they caused and it didn’t flood back, it just dropped traffic within their network.
I wonder if there is anyone at DFA that knows spanning tree??? Good idea to have broadcast rate limiting as a fail safe. Mikrotiks and UBNT implementations are s h 1 t !! Often the wireshark's just don't make sense.
I would say that the best implementation is the stock Linux one (about as good as the Madge implementation-but that one had an awesome visualization tool for spanning tree that ran in DOS which I've never seen before in any of this new fancy pants equipment). There is a tool that is close called EtherApe. Load on a Linux box and connect via a tap or mirror port.
 
Dropped again todayOpenserve Randburg,40mins and counting
 
Top
Sign up to the MyBroadband newsletter
X