29 comments

[ 2.5 ms ] story [ 67.3 ms ] thread
I just saw this as well. Got a downtime alert, but app seems to be working from my location.
Same here. Not sure if Pingdom is acting up or they legitimately can't reach my site due to the problems reported by Cloudflare or something.
Experienced uncharacteristic connection issues in the last ~15 minutes, was primarily SSHing into AWS EC2 at the time though so can’t comment on the true scale.
From cloudflare status [1] posted on HN [2]:

> Network Performance Issues in multiple locations

> The issue is related to a specific transit provider and we are working on temporarily disabling this provider to route around the issue

[1] https://www.cloudflarestatus.com/?

[2] https://news.ycombinator.com/item?id=14246888

That cloudflare status is very strange: it says (said) Angola, Dubai, Philadelphia (US), some Australian locations and some in South America were affected. Anyone have a good explanation of such a random distribution?
I think they're just listing the POPs that were unreachable from their main office. It was definitely more widespread than that.
Telia, the ISP that cause this issue, is very very large -- one of few Tier 1 ISPs. That said, when Tier 1 ISPs experience issues they're incredibly wide spread and behave oddly.

The POPs you mention are the Cloudflare POPs that experienced degraded or dropped throughput due to Telia being (one of?) Cloudflare's preferred routes for that POP.

HN has been sketchy for me for the last half hour or so.
Is it possible to determine the transit provider at fault based on Cloudflare's list of affected areas?
I don't believe that would be possible... (Caveat: I am not the World's Smartest & Bestest Network Engineer Ever™, would be great if other's chimed in!)

1. Each Cloudflare POP has multiple bandwidth providers. 2. Each Cloudflare POP can and does change it's preferred BGP routes. 3. Telia's outage wasn't 100% of their network, so there are plenty of Telia links -- including ones to Cloudflare POPs -- that were not impacted.

OT, but that yellow/white combination of background and font color is terrible.
Looks like telia was down and has fixed a mistake in their configs.
It looks like Telia is apparently the issue.

Our CDN's are having problems all over the place. No indications of what shit the bed, but this is more then Cloudflare or Level3.

Do you have a source? (Would like to pass it on internally)
Multiple ISPs I work with have explicitly stated to me that it was Telia. I'm not aware of any public sources confirming this yet.
No, I have no public sources I can share. Just relaying what our CDN's have shared with me. I also can't provide anything specific. Sorry, I know it's unreliable and rumor but it's all I can provide.
(comment deleted)
I'm curious what Pingdom uses to detect their own outages.
I noticed reddit went down briefly.
I've had Pingdom for about three months now. They have failed to alert on every outage we've had since I started INCLUDING the first "test outage" I created on day 1 to verify the monitoring was working. I am not impressed at all by their service and will probably be asking for a full refund depending on the outcome of my current trouble ticket with them. Also they've been deleting history of our outages after the fact; i.e. an outage occurs, they fail to alert but they do log it, and then several days later the outage is removed from their history. Pingdom is run by absolute clowns.
A BGP problem again? Or something else this time? It amazes me that there has not been an alternative to handle routing except full trust for everyone.
We dropped Telia as a carrier about 2 months ago. I've never seen a network with such stability issues; outages/slowdowns occurred on a weekly basis for us and tech support was useless. The only good thing about Telia is they did a great job with service credits. According to VP's there they have been going through major transitions in the last year but given their Tier-1 status, you would expect more stability.