High packet loss on Vodafone internet

Forums › One New Zealand (including Vodafone, WxC, Farmside) › High packet loss on Vodafone internet

1 | 2

Jase2985

13462 posts

Uber Geek

ID Verified

Lifetime subscriber

#2103974 9-Oct-2018 15:39

I would suggest either number 2 or 3 on the post above. ie the first address that its not inside your network

it could be outside vodafones network, but they want you to test your side of it first as its pretty quick and easy to do.

"We monitor across a bunch of connections and aren't seeing the same behaviour you've picked up on."

"At the moment all we know is you've got issues to a couple of places on the internet - that could be an issue in the wider Vodafone network, your connection specifically, or your device/test. Testing to your router and the next hop along will help narrow this down."

testing to the first server (BGN) will likely eliminate a couple of the places the fault could be.

so try that and report back

gaddman

224 posts

Master Geek

Trusted

#2103977 9-Oct-2018 15:41

yitz: You should be testing the second hop UNASSIGNED.static.cust.vf.net.nz [] I doubt that is your static IP so you won't be testing to your own router, it might be in the same range.

Yeah, that's the BNG. As a rule of thumb, don't trust the DNS entries for our network.

Foiler

220 posts

Master Geek

#2103992 9-Oct-2018 16:14

Ok just got a 6% packet loss reading to the IP address of the first hop UNASSIGNED.static.cust.vf.net.nz

Does this perhaps imply a dodgy router or a dodgy ONT?

Here are the latency readings past 100 days for Google 8.8.8.8 and sip.talk.co.nz. As you can see no long term changes since the onset of packet losses 27-8-18. However, a shorter term view does show big daily variations (40/160msec cycles) of Google latency since 3-10-18 which may be what @noroad was referring to.

Foiler

220 posts

Master Geek

#2104024 9-Oct-2018 16:53

Have now added a PRTG sensor for the WAN address of the Draytek router and that is also showing packet losses.

Will do some more testing, firmware upgrade etc etc and to close off this case I will advise when I have more info and a conclusion.

noroad

951 posts

Ultimate Geek

Trusted

#2104260 10-Oct-2018 08:46

I would suggest there are two things going on there, firstly the latency changes which are just Google routing. Secondly the packet loss which is either the local router, UFB connection or the Vodafone backhaul between the LFC and their BNG (or the BNG itself).

Foiler

220 posts

Master Geek

#2104266 10-Oct-2018 09:07

Thanks noroad .. I think the packet losses measured from LAN to WAN of the router suggest that the losses are actually in the Draytek router. Now doing further tests to ascertain the reason for this.

noroad

951 posts

Ultimate Geek

Trusted

#2104274 10-Oct-2018 09:21

Foiler:

Thanks noroad .. I think the packet losses measured from LAN to WAN of the router suggest that the losses are actually in the Draytek router. Now doing further tests to ascertain the reason for this.

Normally a company with an extensive network such as Vodafone will have in-depth monitoring of the network that will tell them if there are loss issued on the backhaul/BNG path. But, as we have all seen with the HFC network fiasco in times past this does not always guarantee its not on the service provider end. I used to use Draytek myself before upgrading to gig fibre and think they are great devices, but I found performance over a couple of hundred megs did not really stack up once the load came on.

Sharesies

Trade NZ and US shares and funds with Sharesies (affiliate link).

gaddman

224 posts

Master Geek

Trusted

#2105571 10-Oct-2018 13:17

Foiler:

Thanks noroad .. I think the packet losses measured from LAN to WAN of the router suggest that the losses are actually in the Draytek router. Now doing further tests to ascertain the reason for this.

Just check your path/latency - some home routers will send all packets into the WAN network first, so the you may be testing your upstream link at the same time (rather than just testing your router). Also, it could be your router's a bit iffy about responding to pings. Do you see any loss pinging the router's LAN interface?

Foiler

220 posts

Master Geek

#2105620 10-Oct-2018 14:32

Packet loss to LAN IP of router is nil, 0msec.

Occasional 1% - 3% packet loss to WAN IP of router, 1msec. Pretty sure this is just the router and no network beyond.

Ping Jitter to the WAN ip is only 0.5 constant all day. By comparison Ping Jitter to Vodafone fist hop mostly 1-10 but occasional jumps to 45 when there is high throughput.

Any suggestions for SNMP traps? I am now monitoring LAN and WAN throughput, Unicasts etc ..

Foiler

220 posts

Master Geek

#2115730 29-Oct-2018 08:01

Have replaced the router with newer model Draytek 2862 but packet losses are persisting, though to a much lesser degree and less often.The original Draytek 2860 has been thoroughly tested by the supplier and no fault found - no packet losses at all under their test setup.

Interestingly another site with a Draytek 2860 but connected to Spark by VDSL is also showing packet losses when pinging the WAN address of the Draytek from the LAN side.

The SNMP traps displaying LAN and WAN traffic do not seem to be showing any excessive activity, nor any correlation between traffic and packet losses.

Anyone got any further ideas on this or what other tests to run in order to solve this one?

The overriding puzzle for me is that the packet losses seemed to start suddenly on 27-8-18 after a year of hardly a packet loss .. so is this behaviour possibly a Chorus feature? unlikely given that it is showing on both UFB and VDSL and with two different ISPs?

gaddman

224 posts

Master Geek

Trusted

#2116019 29-Oct-2018 11:58

I'm not too surprised you're getting packet loss pinging the WAN IP - that's probably just your router not caring to respond.

I wonder if it could be your test method that's indicating the issue, that may explain why you saw a jump to both destinations at the same time. That's assuming some update or other change at that time, but even changing the number of packets could cause that - a ping every RTT is low impact but 100 pings in short succession could trigger something to discard your packets. Do you see the same packet loss testing from a different PC using command line ping (rather than PRTG)? Also with command line you might find it easier to see when the PL occurs, eg

$ ping -D 121.74.15.254
PING 121.74.15.254 (121.74.15.254) 56(84) bytes of data.
[1540767275.517810] 64 bytes from 121.74.15.254: icmp_seq=1 ttl=254 time=5.85 ms
[1540767276.521153] 64 bytes from 121.74.15.254: icmp_seq=2 ttl=254 time=8.08 ms
[1540767277.524191] 64 bytes from 121.74.15.254: icmp_seq=3 ttl=254 time=9.80 ms

1 | 2