We're seeing intermittent network outages for our Dallas and Ashburn network due to issues with CosmicGuard's network, they're working on a resolution.
Issues have started again, CosmicGuard has said they have a plan to resolve the disruption, caused by large attacks leaking past Cloudflare.
We haven't seen any disruption for the past 35 minutes so it's likely this has now been resolved.
CosmicGuard has now updated their status page: https://status.as30456.net/en-gb/cmhhpokrs00s1omomxdm6ppa4
CosmicGuard has confirmed they're actively working on a resolution with Cloudflare to stabilise the network.
The network instability has been continuing, with disruptions every ~5 minutes affecting almost the whole of CosmicGuard's global network. Our current assumption is that ddos attacks are likely the cause, we haven't yet received an update from CosmicGuard.
Our Seattle DDoS protected network saw instability starting at 20:37 UTC and ending at 20:45 UTC, this was caused by one of our DDoS mitigation providers, Neoprotect, suffering a full network outage, our backup network automatically took over within minutes and this will remain active until Neoprotect is stable again.
See the update below for full details:
Neoprotect has stated that they are not going to continue their service. We are investigating alternative DDoS mitigation providers and hope to have one available soon, this will also return latency to normal levels.
Neoprotect is still offline, they use datapacket/CDN77 as a sole upstream and they disabled Neoprotect's service which resulted in the full global outage of Neoprotect's network. You can read more about the cause here: https://status.neoprotect.net
We are currently routing DDoS protected inbound traffic through Cnservers Los Angeles so there will be a latency increase for locations that are geographically close to Seattle, if this is a problem for you in the short term, please contact us and we'll discuss the options available with you.
Longer term, we will onboard another ddos mitigation provider, potentially GSL, that has a local Seattle POP, this will ensure latency is normal.
Our DDoS protected Dallas network is currently offline due to an issue on CosmicGuard's side, we are talking to them.
It has since become known that Cloudflare is currently performing large-scale maintenance in many of their locations, their Dallas location went completely offline for the period we were offline, so this explains why we saw an outage. There was likely something that went wrong on Cloudflare's side in another of their locations that somehow then impacted Dallas, CosmicGuard is setup to automatically fail us over to another of their locations in the event of BGP routes being pulled - something that Cloudflare did not do in Dallas for an unknown reason. CosmicGuard will be talking to Cloudflare about this in the hope of avoiding a repeat.
Dallas came back online at 08:40 UTC. CosmicGuard has been using Cloudflare Magic Transit recently to help absorb the large attacks that DDoS mitigation providers are being hit with nowadays, 20Tb/s+. Unfortunately Cloudflare suffered a connectivity issue to CosmicGuard in Dallas, which resulted in traffic being blackholed, essentially dropped. CosmicGuard is currently investigating with Cloudflare why this happened, to hopefully prevent a repeat. We apologise for the inconvenience caused.
No incidents reported
We're currently seeing increased latency to/from China for our Seattle network due to suspected fiber cuts on the US west coast, this appears to be heavily impacting China Unicom's network (AS4837) which has likely lost considerable capacity. CMIN2 should not see any impact apart from a temporary 20ms latency increase due to a re-route caused by the fiber cuts. This shouldn't take too long to be resolved.
The fiber cut was fixed at 4pm UTC and we can confirm everything is now looking normal.
Confirmed fiber cut, here is the explanation from our transport provider: There is a fiber cut between Vacaville, CA and Sacramento, CA. The fiber cut was caused by a car hitting a utility pole.
It's likely that China Unicom is suffering capacity loss due to this same incident, there is no estimated fix time yet.
Our Seattle DDoS protected network is currently seeing high packet loss, we're investigating.
This appears to be resolved, the issue was related to firewalls upstream of our network. We're working on changing the route used for non China traffic to make it much more stable for the future, this should be completed over the next 5 days at most. For China traffic, we're working with Ceranetworks to ensure stability and are also looking further into different China optimised routes we can use.
No incidents reported
No incidents reported
No incidents reported
No incidents reported