Web Squad ISP

Status
Not open for further replies.
I'm in Durban and it's down
Cpt Southern suburbs down

And everyone else who commented.

Thank you all for your patience. We brought down some services in CPT a little earlier to complete an upgrade at the DC - which should have been hitless (besides a small latency bump during the upgrade). However, we didn't anticipate some routes to behave the way they did (some advertisements didn't withdraw correctly leaving some traffic blackholed (ie with nowhere to go)). This included some routes in Durban which were sending traffic to Cape Town. We'll definitely look at the mistakes here going forward to ensure it doesn't happen again.
 
@websquadza - something going on? Stability really seems to have tanked the last few days.

Cape Town was mostly unaffected by Joburg's issue this week. Unfortunately, your Google issues were more widespread than just our network, as there was a capacity issue on their Joburg 1 Core (This also feeds JINX, which we call Google 3). This caused issues with some services on certain devices. We drained their JHB 1 late yesterday pending their maintenance and upgrades overnight. Google is completely back up now and we're monitoring this carefully.
 
Something going wrong on my side, can't load news24.com, getting the below time out error each time and constant disconnects from COD Modern Warfare games. I hope my IP is not again getting blacklisted for packets:

1590166265730.png
 
Something going wrong on my side, can't load news24.com, getting the below time out error each time and constant disconnects from COD Modern Warfare games. I hope my IP is not again getting blacklisted for packets:

View attachment 844495
News24.com seems down in general. Just tried via mobile (Voda) and can't access it either.

COD may have been a final reconvergence our side. We just pushed one last routing adjustment 10 minutes ago and took a min or so to come together.. Don't worry - not breaking/touching things further tonight. Just keeping an eye to make sure all the changes/upgrades do what they're meant to.
 
Hey @websquadza

Just a quick report to let you know that my google news' buffering issues with the 1st skynews clip has not happend for a good few days. If you did something, thanks; if you didn't, thanks anyways :)
 
Cape Town was mostly unaffected by Joburg's issue this week. Unfortunately, your Google issues were more widespread than just our network, as there was a capacity issue on their Joburg 1 Core (This also feeds JINX, which we call Google 3). This caused issues with some services on certain devices. We drained their JHB 1 late yesterday pending their maintenance and upgrades overnight. Google is completely back up now and we're monitoring this carefully.

Just wanted to check if this was fully resolved? I still get a streaming experience specifically on iPhone which is not what it used to be. Confident it’s not my local network, Speedtest reports plenty of bandwidth, but seemingly YouTube is not happy, it’ll sometimes play fine and other times the quality is low.

In times where the quality is low, anything over 360p won’t play. However oddly enough, if I let the video play through, close and reload the video and choose another quality, (eg 1080p) it’ll then play perfectly with plenty of buffer-ahead.

I am assuming there is something with a cache or the content coming down from different sources in this instance somewhere.

There also seems to be a persistent pattern where it’ll play for 10 seconds nearly exactly, then pause and buffer for quite a while, and then resume and play just fine.

Conversely, from an Android TV device on the network, it appears to buffer all the way up to 4K consistently no problem.

I know this is quite specific and probably very hard to troubleshoot, so I understand if everything is deemed normal otherwise. Just wanted to give as much info as possible.

Speedtest from my phone attached just for prosperity sake.

aadbf3b5cc62ea32f1c246ba436855ea.jpg
 
Hi @websquadza just checking to see if the openserve (AS5713) routing has been resolved?


Also see seacom (AS37100) is way off too


Seacom have been, well Seacom - since they de-peered everyone last year and want a fair whack of money for transit traffic - basically forcing ISPs to use their transit if they want to exchange traffic with them - exchanging traffic has been hard..

SAIX woke up to this tactic again last week (after slowly releasing a few prefixes and accepting from the exchanges over the past while) and withdrew advertisements of all prefixes that aren't Telkom ISP specific (these they've kept up for the time being). A few emails to SAIX/Openserve/Telkom NOC later and an Openserve Sales manager has been in touch to try to convince us that their transit is the way to go (note this is the same transit that failed them and some other big ISPs twice during recent undersea outages). So basically, two incumbents have resorted to extreme measures to sign up new wholesale business. This while we all sit on at least 5 mutual exchanges and peer with 400+ other networks, for free.

That said, it does affect our clients when the quickest route to something down the road is via London. We're exploring options at the moment as part of our transit upgrades and will have something more permanent in place soon.
 
Just wanted to check if this was fully resolved? I still get a streaming experience specifically on iPhone which is not what it used to be. Confident it’s not my local network, Speedtest reports plenty of bandwidth, but seemingly YouTube is not happy, it’ll sometimes play fine and other times the quality is low.

In times where the quality is low, anything over 360p won’t play. However oddly enough, if I let the video play through, close and reload the video and choose another quality, (eg 1080p) it’ll then play perfectly with plenty of buffer-ahead.

I am assuming there is something with a cache or the content coming down from different sources in this instance somewhere.

There also seems to be a persistent pattern where it’ll play for 10 seconds nearly exactly, then pause and buffer for quite a while, and then resume and play just fine.

Conversely, from an Android TV device on the network, it appears to buffer all the way up to 4K consistently no problem.

I know this is quite specific and probably very hard to troubleshoot, so I understand if everything is deemed normal otherwise. Just wanted to give as much info as possible.

Speedtest from my phone attached just for prosperity sake.

aadbf3b5cc62ea32f1c246ba436855ea.jpg

Thanks for update here. What I can say is that Google JHB caches (1,2 and 3) are back down to delivering almost no traffic since Sunday mid day. Their "repairs" in JHB gave in after 48 hours. This means that traffic tries to reach you from anywhere else, including our KZN cache and international transit - and Google's network is not designed to run over transit (ie high latency). We've reached out to their NOC again and @cobusv has been great at providing feedback from the JHB side. We'll keep you updated as we hear more here.
 
Thanks for update here. What I can say is that Google JHB caches (1,2 and 3) are back down to delivering almost no traffic since Sunday mid day. Their "repairs" in JHB gave in after 48 hours. This means that traffic tries to reach you from anywhere else, including our KZN cache and international transit - and Google's network is not designed to run over transit (ie high latency). We've reached out to their NOC again and @cobusv has been great at providing feedback from the JHB side. We'll keep you updated as we hear more here.

Thanks very much for the info. This makes sense. Appreciate the feedback very much.
 
Seacom have been, well Seacom - since they de-peered everyone last year and want a fair whack of money for transit traffic - basically forcing ISPs to use their transit if they want to exchange traffic with them - exchanging traffic has been hard..

SAIX woke up to this tactic again last week (after slowly releasing a few prefixes and accepting from the exchanges over the past while) and withdrew advertisements of all prefixes that aren't Telkom ISP specific (these they've kept up for the time being). A few emails to SAIX/Openserve/Telkom NOC later and an Openserve Sales manager has been in touch to try to convince us that their transit is the way to go (note this is the same transit that failed them and some other big ISPs twice during recent undersea outages). So basically, two incumbents have resorted to extreme measures to sign up new wholesale business. This while we all sit on at least 5 mutual exchanges and peer with 400+ other networks, for free.

That said, it does affect our clients when the quickest route to something down the road is via London. We're exploring options at the moment as part of our transit upgrades and will have something more permanent in place soon.


Woot no way. That is a very bad way of doing business wow. Very swak move from Telkom there.
 
@websquadza not sure if I am seeing ICMP filtering, but seeing some pretty bad packet loss to 1.1.1.1 and 9.9.9.9 which appears to be both the DNS servers that DHCP provides me.
Definitely making things feel very slugglish tonight, resolving pages etc times out and then retries and succeeds frequently. Will try 8.8.8.8 instead as a test.
Tests from gateway...
 

Attachments

  • Screenshot 2020-05-26 at 20.59.27.png
    Screenshot 2020-05-26 at 20.59.27.png
    326.1 KB · Views: 24
Status
Not open for further replies.
Top
Sign up to the MyBroadband newsletter
X