Hmm, all gone poor again:
Pings are up too. Is the 185 range OK or should I be put on to the 141 range?
Cheers.
Keeop
1 Like
Iād prefer the 185 range pleaseā¦ This one is shit
Iām losing my patience now with this serviceā¦
Can you try a few different speedtest servers for me?
Weāve seen some crap upload speeds with M247 over the last day or so, sometimes even with non Yayzi connections
Let me know your preference and Iāll post results here
Still seeing slow speeds here - Iāve forced a couple of clients to go via Yayzi while the rest of the network remains on Virgin and from the wired client:
$ speedtest -s 30690
Speedtest by Ookla
Server: Community Fibre Limited - London (id: 30690)
ISP: Yayzi Broadband
Idle Latency: 5.98 ms (jitter: 0.07ms, low: 5.92ms, high: 6.04ms)
Download: 12.52 Mbps (data used: 14.4 MB)
567.38 ms (jitter: 89.15ms, low: 23.87ms, high: 1375.26ms)
Upload: 0.33 Mbps (data used: 186.4 kB)
2883.78 ms (jitter: 226.19ms, low: 857.88ms, high: 3450.99ms)
Packet Loss: 0.0%
Result URL: https://www.speedtest.net/result/c/b703b9a1-f5c1-4193-b205-b77957b502b3
$ curl ipinfo.io
{
"ip": "185.210.170.129",
WiFi client:
It doesnāt seem to matter which server I choose, or which speedtest, the result is pretty awful in terms of speed. Any thoughts, @Yayzi_Team ?
Are you using our router or your own? This is certainly not right
I had complaints from my son about Fortnite last night as I stuck him on the Yayzi connection for him to test. Strange, as I was monitoring everything and the latency/jitter etc. seemed fine so am assuming itās just more to do with the route to the servers. He said it was a bit āglitchyā which I assume means laggy. He said that there was some packet loss displayed on the in game overlay along with some symbols that came up a few times indicating a dropped or weak connection. Heās fully wired so was not internal issue. I stuck him back on VM and he said it was then all OK so something there for you to look at please.
When heās playing different stuff, will try and do the same thing again.
Cheers.
Keeop
My own opnsense box. Speedtest via the Virgin gateway (on said opnsense box) is what Iād expect (about 7-800Mbit down, 110Mbit up) but via the Yayzi gateway itās awful (though it doesnāt feel as bad as it looks when browsing) - FWIW this worked fine prior to the great outage, I got right around line speed; itās even worked fine at points during all the reorganisation work, but not for the last few daysā¦
Just looking at the thinkbroadband monitor, you can see as soon as I start āexercisingā the connection thereās significant packet loss reported:
1 Like
Forgive me for butting in, but sanity checking with the supplied Yayzi router for 5 minutes is probably the easiest way to move this forward as it removes any possibility of it being an OPN quirk and puts it back to a Yayzi issue (sorry Liam). Also I fully appreciate the irony/humour in you having to use a dynamic dns provider for a static connection, but possibly better to generate an anonymised BQM in future.
Yeah, itāll probably be some time before I can put the Yayzi router back in and do any testing (definitely not during the work day, anyway) - and this configuration was working just fine prior to the outage.
Iām not too worried about the DNS name being there - a) I donāt think having half a dozen IPs since signing up a week ago really counts as static and b) we all know the DNS will have been enumerated within minutes of being created, and a thousand different mindless botnets will be scanning this and every other host on the internet for holes 
Well this is ironic. Random power cut last night (donāt worry Liam, you arenāt getting the blame for that 
), servers shut down as expected via UPS, and I left everything off over night as I have some maintenance planned for today, and a totally idle BQM graph would be useful at this point as you can literally see the hourly spikes when my speedtest server checks the connection. Now I know I joked above about the humour in having to run a dynamic DNS service for a connection with a static IP, but I shouldnāt have, as my IP changed and no BQM, so today iām setting up DDNS on my static connection 
@aaronjb did you manage to get swapped out and confirm either way?
Look, if I learnt one thing in my time hereā¦ is that we can blame @Yayzi_Liam for everything. Right!?
Speaking of speeds - I did; I swapped the EX820v in last night and did a quick test (over the stock wifi because I picked the wrong laptop up ā¦ give me back ethernet ports on laptops, manufacturers! shakes fist at Asus and Dell ā¦ anyway) and speeds were great. I mean, 6-700Mbit down and 4-500Mbit up; thatās a lot less than line rate but that could easily be down to testing over WiFi.
So now Iām off down an opnsense rabbit hole. Still weird, because I definitely used to get line rate via the opnsense box; of course that was before I went dual-WAN and added firewall rules for inbound ICMP. Neither should have any impact but something certainly is! So Iām staring at tcpdumpsā¦
ā¦itās like being at work 
Right. Yes. So.
Iām a moron.
# speedtest
Speedtest by Ookla
Server: YouFibre - London (id: 55137)
ISP: Yayzi Broadband
Idle Latency: 3.48 ms (jitter: 0.06ms, low: 3.45ms, high: 3.55ms)
Download: 1218.91 Mbps (data used: 609.9 MB)
8.46 ms (jitter: 0.48ms, low: 3.46ms, high: 9.45ms)
Upload: 968.65 Mbps (data used: 1.0 GB)
7.92 ms (jitter: 0.74ms, low: 3.82ms, high: 9.79ms)
Packet Loss: Not available.
Result URL: https://www.speedtest.net/result/c/fb0ab80f-e347-4e22-bf67-b00bb2555a06
At some point I had futzed with the priority code point for the Yayzi VLANā¦ maybe as far back as when all the line instability was there (going up/down etc) and Yayzi were struggling with DHCP issues. Set it back to 0 and voila, all my problems disappeared.
Itās a good job my job title doesnāt have āNetwork Engineerā in it anymoreā¦
3 Likes
Ah, the good old picnic error. Weāve all been there!
1 Like
Weāve all done it at some point and fair play for the admission, itās more than many are prepared to make. Iām sure @Yayzi_Team wonāt hold any grudges, just look out for a dartboard in the background of a future blog photo with some familiar looking faces/avatarās in key positions
Interesting. My IP hasnāt changed but all the Speedtest server locations that are selectable have - I can no longer select the Community London server and what few there are seem to be in the Midlands now! Looks like Iām quite near Coventry, apparently.
Not sure whatās going on at Yayzi HQ today thenā¦
Cheers.
Keeop
I think $dayjobās Product Engineering team probably have a voodoo doll of me that Yayzi can borrow, as well
@Keeop , are you still on 185.x? (I am and I see consistent speeds now Iāve fixed my slight errorā¦)
@aaronjb Not consistent, no. Getting between 1200 and 1800 mostly, but my IP location seems to have changed as the servers itās finding for me are all in the Midlands which is not where I am based. I can still choose London - I wasnāt seeing the option on the screen I was on earlier which was my mistake - but it defaults to a load of servers in the Midlands. If I choose Community Fibre London, then I am getting better speeds so will have to run some more tests and make sure I have that one selected each time.
OK, so just run a load over the last 10 mins for consistency. Only got the one machine on the Yayzi at the moment, and itās only running these tests. So, I had a 700Mb difference between fastest and slowest. Still pretty wild variance.
Hmm, just ran another and got around 1100Mb whereas the best I got earlier on was 2020Mb. Also, on this last test, the latency really took a hit - sort of thing that happens with VM when there are too many subscribers on the network segment. Run again, 1450 with increased latency but not as bad as previous. Whole thing just seems all over the place. Iām not really complaining as those speeds are more than enough but would be nice to know whatās causing such variation.
Maybe itās just how it is? Just a more noticeable difference with a faster connection?
1 Like
I wouldnāt say this seems to be correct. The network is only using about 12% capacity right now.
Iād say there is a good chance the speedtest servers are possibly under load as well. But I donāt want to come across as weāre kinda shifting the blame there either.
But certainly no traffic shaping or anything from our end