Web Squad ISP

Status
Not open for further replies.
Thanks for this
MTR looks clear. Definitely not imagining things, just trying to isolate the potential cause. Desktop youtube loading fine? Do you have IPv6 enabled on your router? Also try swapping DNS to cloudflare 1.1.1.1
Desktop youtube seems fine, don't have IPv6 enabled, never gotten around to it.
Swapping DNS on mobile to 1.1.1.1 didn't make a difference to loading,
Have swapped desktop to 1.1.1.1 also to see if the Google meets break ups go away.
 
Desktop youtube seems fine, don't have IPv6 enabled, never gotten around to it.
Swapping DNS on mobile to 1.1.1.1 didn't make a difference to loading,
Have swapped desktop to 1.1.1.1 also to see if the Google meets break ups go away.
Oh god, Google Meets have been a nightmare for a couple of weeks with random stoppages every now and again, 1.1.1.1 won't help with those, I'm using CF for DNS as it is. We (company) suspect that something iffy happened on Google's side, because there are only issues when we are in a meet with people from EU/UK. US/Asia/Africa is fine.
 
Oh god, Google Meets have been a nightmare for a couple of weeks with random stoppages every now and again, 1.1.1.1 won't help with those, I'm using CF for DNS as it is. We (company) suspect that something iffy happened on Google's side, because there are only issues when we are in a meet with people from EU/UK. US/Asia/Africa is fine.
I was just about to set up a test call to an offnet server and see how long it lasts, but you reckon it's more the international leg?
 
Desktop youtube seems fine, don't have IPv6 enabled, never gotten around to it.
Swapping DNS on mobile to 1.1.1.1 didn't make a difference to loading,
Have swapped desktop to 1.1.1.1 also to see if the Google meets break ups go away.
What router are you using?
 
I was just about to set up a test call to an offnet server and see how long it lasts, but you reckon it's more the international leg?
It's something in the EU. Spent a good 3 hours with infrastructure guys from Canada last week with the only drop being loadshedding switchover. A few hours earlier we had the CTO in the UK on the same call and was dropping (Canadian guys were also dropping) every 20ish minutes. Moving to teams this week because of it.
 
Oh god, Google Meets have been a nightmare for a couple of weeks with random stoppages every now and again, 1.1.1.1 won't help with those, I'm using CF for DNS as it is. We (company) suspect that something iffy happened on Google's side, because there are only issues when we are in a meet with people from EU/UK. US/Asia/Africa is fine.
Follow up here, any luck on those tests?
 
It's something in the EU. Spent a good 3 hours with infrastructure guys from Canada last week with the only drop being loadshedding switchover. A few hours earlier we had the CTO in the UK on the same call and was dropping (Canadian guys were also dropping) every 20ish minutes
If there are error outputs on Google meets (similar to Youtube), please forward these and I'll post a ticket on the Google ISP portal. This sounds like a large issue on their internal routes?
 
If there are error outputs on Google meets (similar to Youtube), please forward these and I'll post a ticket on the Google ISP portal. This sounds like a large issue on their internal routes?
Nothing, diagnostics only shows a slight ms bump (20-30ms) before a huge stoppage. Got a 2 hour call with the infrastructure guys in the UAE in 20 minutes, so will keep an eye on it again.
 
LON1 is stable again. Issues are so transitory. Wonder if it is SFP related. I just don't have time to come out and collect a new one from you right now :sob:

1676280872493.png
 
Nothing, diagnostics only shows a slight ms bump (20-30ms) before a huge stoppage. Got a 2 hour call with the infrastructure guys in the UAE in 20 minutes, so will keep an eye on it again.
Please do, Google NOC is painful, so any and all stats at call fail will help.

LON1 is stable again. Issues are so transitory. Wonder if it is SFP related. I just don't have time to come out and collect a new one from you right now :sob:

View attachment 1474915
Thanks for confirming. Hop 5 is the handover to us (2,3,4 are our IPs configured in a VRF within Vuma's core network, with 2 being your area L3 headend, and 3 and 4 being internal routing before reaching us on 5. 3 and 4 are severely ICMP rate limited), and in your previous plot, 5 showed almost 50% drops. We do also implement ICMP rate limiting on certain devices, but not 50%. Obviously with first hop not being clear, we can't see if that SFP in the CPE was at play today.
 
The YouTube issues are back and I confirmed a VPN resolves it. I shared an update in my ticket.

The issue seems to be specific to YouTube (or Google) since other websites are working
 
The YouTube issues are back and I confirmed a VPN resolves it. I shared an update in my ticket.

The issue seems to be specific to YouTube (or Google) since other websites are working
We do think this has something to do with Google peering in JHB, but trying to isolate it. VPN usually helps for one of two reasons:
1. MTU issues: VPNs auto clamp MSS for connections, so they resolve MTU issues
2. Bypass local routes: VPNs tend to carry traffic from elsewhere, so might be that Google has an issue on JHB peering, and the VPN is picking up the traffic elsewhere.
 
The YouTube issues are back and I confirmed a VPN resolves it. I shared an update in my ticket.

The issue seems to be specific to YouTube (or Google) since other websites are working
Desktop youtube seems fine, don't have IPv6 enabled, never gotten around to it.
Swapping DNS on mobile to 1.1.1.1 didn't make a difference to loading,
Have swapped desktop to 1.1.1.1 also to see if the Google meets break ups go away.

Can you check on this again? We've updated announces and trying to steer to different Google egress ports to see if it makes a difference
 
Please do, Google NOC is painful, so any and all stats at call fail will help.


Thanks for confirming. Hop 5 is the handover to us (2,3,4 are our IPs configured in a VRF within Vuma's core network, with 2 being your area L3 headend, and 3 and 4 being internal routing before reaching us on 5. 3 and 4 are severely ICMP rate limited), and in your previous plot, 5 showed almost 50% drops. We do also implement ICMP rate limiting on certain devices, but not 50%. Obviously with first hop not being clear, we can't see if that SFP in the CPE was at play today.
I'll disable ICMP filtering on the router for a few days tonight.
 
Can you check on this again? We've updated announces and trying to steer to different Google egress ports to see if it makes a difference
I noticed now that my YouTube app was still giving problems, but opening YouTube in a browser was working.

Update: issue is back
 
Status
Not open for further replies.
Top
Sign up to the MyBroadband newsletter
X