Spotify regularly pausing

Also as a solution , on windows you can use the hosts file and block these IPs (mweb) and Spotify should reroute accordingly solving all the issues here

Thanks for that tip @cavedog
 
I brought this up a week or two ago in the Cool Ideas thread. Using their DNS didn't help. Seems like Akamai was just doing its own thing. Would hop between Mweb, International and some other random servers.
Yerp.
 
I tracked down one of the DNS names spotify uses: a711.dscd.akamai.net

Found the IP in resource monitor:
View attachment 1597790

Then ran: ipconfig /displaydns | clip
Then pasted results into Notepad and found the DNS name:

View attachment 1597794

I then compared CISP vs Afrihost's DNS and it looks like Afrihost's returns the on-net IP for their cache, but every other DNS server I try returns MWEB.

View attachment 1597798

Edit: CloudFlare DNS is returning a direct akamai entry, might be worth trying 1.1.1.1:

View attachment 1597802
I appreciate the effort.

I tried switching to 1.1.1.1, did an `ipconfig /flushdns`, restarted Spotify, and still connecting to MWeb IP. I tried `ipconfig /displaydns | clip`, but the results never contain that IP (only the reverse lookup). I even tried re-installing Spotify again just in case.
 
Also as a solution , on windows you can use the hosts file and block these IPs (mweb) and Spotify should reroute accordingly solving all the issues here

Thanks for that tip @cavedog
Do you have a link to the exact post so I can see which hosts to block in my file?
 
CDNs have a bit of black magic algorithms to them, which should be basically. On-net, next net, peer. Or just availability and capacity metrics.

So who knows how they manage it. Our cluster has loads of capacity and is on net, so why not use it?
 
I was just going to quote cavedog's post as well, @PBCool according to @cavedog you guys can request to unmap optinet.

I am seeing your DNS servers return MWEB/Optinet CDN IPs for domains that used to return on-net IPs in the past.

View attachment 1597522

Yea i saw the post late last night and was going to ask so when i saw the thread i was like OH YES I MUST ASK :ROFL:

Looks like I have some followers :unsure: :ROFL:

I don't know what is going on at Optinet and their commercial agreement Akamai but it's horrible. What is even worse is they advertise those cache to all ISPs at NAP JHB, DUR and CPT which means if you are in KZN for example and get a CPT Optinet cache your ISP will hand over the traffic to Mweb at NAP DUIR and they will transit it to CPT via JHB so congested NLD now with Webafrica on board just makes things worse.

I had to fight Akamai a bit for Afrihost but after a day or so they agreed and removed the Optinet caches from our ranges. It's not served on-net or peering. After that complaints just went away.
 
Looks like I have some followers :unsure: :ROFL:

I don't know what is going on at Optinet and their commercial agreement Akamai but it's horrible. What is even worse is they advertise those cache to all ISPs at NAP JHB, DUR and CPT which means if you are in KZN for example and get a CPT Optinet cache your ISP will hand over the traffic to Mweb at NAP DUIR and they will transit it to CPT via JHB so congested NLD now with Webafrica on board just makes things worse.

I had to fight Akamai a bit for Afrihost but after a day or so they agreed and removed the Optinet caches from our ranges. It's not served on-net or peering. After that complaints just went away.
Also, word on the street is Mweb throttles the transit to that cluster due to the amount of transit it has been using. So makes sense.
 
Looks like I have some followers :unsure: :ROFL:

I don't know what is going on at Optinet and their commercial agreement Akamai but it's horrible. What is even worse is they advertise those cache to all ISPs at NAP JHB, DUR and CPT which means if you are in KZN for example and get a CPT Optinet cache your ISP will hand over the traffic to Mweb at NAP DUIR and they will transit it to CPT via JHB so congested NLD now with Webafrica on board just makes things worse.

I had to fight Akamai a bit for Afrihost but after a day or so they agreed and removed the Optinet caches from our ranges. It's not served on-net or peering. After that complaints just went away.
I've never heard of Optinet before, but from how you're talking about it, it sounds like they're the one with the relationship with Akamai and are using Mweb's infrastructure to host servers used by Akamai.

Possibly they get to bill Akamai per GB served, thus creating this perverse incentive where they convince Akamai they can serve the entire country with little regard for efficient use of infrastructure or that it's creating a terrible experience for Akamai users.

I wouldn't know how much Mweb cares about this, but they may be locked into a contract with Optinet and their only option to limit the impact to their network is by throttling to these particular servers, thus making the experience of users even worse.
 
I've never heard of Optinet before, but from how you're talking about it, it sounds like they're the one with the relationship with Akamai and are using Mweb's infrastructure to host servers used by Akamai.

Possibly they get to bill Akamai per GB served, thus creating this perverse incentive where they convince Akamai they can serve the entire country with little regard for efficient use of infrastructure or that it's creating a terrible experience for Akamai users.

I wouldn't know how much Mweb cares about this, but they may be locked into a contract with Optinet and their only option to limit the impact to their network is by throttling to these particular servers, thus making the experience of users even worse.
Nah, Optinet was the network division of Mweb historically. They were collapsed into DD when they were acquired.

It's possibly one of the first or largest clusters around, why it keeps getting preferred. Who knows.
 
231006-1Q3V9N5V1d.png

Spotify tends to buffer worse at night and on weekends, This is using CISP DNS
It seems audio-ak-spotify-com.akamaized.net which is the cname of a297.dscc.akamai.net. (197.80.131.162) seems to be where the audio stream is coming from.
 
Last edited:
View attachment 1598302

Spotify tends to buffer worse at night and on weekends, This is using CISP DNS
It seems audio-ak-spotify-com.akamaized.net which is the cname of a297.dscc.akamai.net. (197.80.131.162) seems to be where the audio stream is coming from.
Called peak times, right, so it's not a major surprise. Hopefully, get them to steer better with the changes.
 
Nah, Optinet was the network division of Mweb historically. They were collapsed into DD when they were acquired.

It's possibly one of the first or largest clusters around, why it keeps getting preferred. Who knows.
Okay, I see. Possibly just apathy by Mweb/Optinet/Akamai.

In the meantime I've used my CISP issued Archer C20 router to block 197.80.130.0 - 197.80.130.255 which seems to be helping.
 
Ever since the recent cable breakage my Spotify often pauses while playing music.

It seems like when it's struggling to stream that I'm talking to an MWeb server:
View attachment 1597518
So the issue may not be with Cool Ideas.

Everything seems to be working perfectly, for example I do hours of Teams video calls a day without issue.

Is anyone else experiencing Spotify issues? Are there still cables needing to be repaired which might explain why the CDN server is not always keeping up?

Also, I'm connected by cable to my router, not using WiFi.
Using a browser or app? With me I found that it pauses in the browser all the time and never with app (mobile and desktop)
 
Using a browser or app? With me I found that it pauses in the browser all the time and never with app (mobile and desktop)
With the desktop app.

When I tested with the browser version it was generally worse, likely because it can't cache as well locally as the desktop app.
 
Looking good on my side too , interestingly even using 8.8.8.8 or 1.1.1.1 no longer resolves MWeb's servers, and mildly amusingly, Google resolves to CISP's servers, while CISP's DNS server does not.

1696687854369.png

I suspect Google and Cloudflare pass on the original IP in some way to CDN providers like Akamai so that the "optimal" PoP IP can be resolved.
 
Looking good on my side too , interestingly even using 8.8.8.8 or 1.1.1.1 no longer resolves MWeb's servers, and mildly amusingly, Google resolves to CISP's servers, while CISP's DNS server does not.

View attachment 1598674

I suspect Google and Cloudflare pass on the original IP in some way to CDN providers like Akamai so that the "optimal" PoP IP can be resolved.

Edge connect.

That is interesting.
 
Top
Sign up to the MyBroadband newsletter