4 ms·
This is the fourth downtime event this week, and it won't be the last. Linode appears to have an abusive relationship with the people at Hurricane Electric. N
by diolpah 15y ago
This is the fourth downtime event this week, and it won't be the last.
Linode appears to have an abusive relationship with the people at Hurricane Electric. No matter how poorly HE treats Linode, Linode always takes it and begs for more.
Our failure management scripts have gone into action 4 times this week because of Fremont, but our relatively high DNS TTL of 15 minutes has still left a few unhappy west coast customers. We will be reducing our TTL and seeking a new west coast provider. Most likely back to ec2.
- nodesocket 15y agoDo you basically run mirrored Linodes in different data centers, and rysnc everything from the primary to secondary? Then when primary goes down, update the DNS to secondary?
- diolpah 15y agoSort of. Our site instances are shared-nothing peers, so there's no primary/secondary. Users resolve the closest host to them ( anycast geoIP )by default, but if one of the hosts goes unavailable, the DNS entry for that host is removed by our scripts.
- nodesocket 15y agoCool, who's your DNS provider who does the anycast geoIP?
- diolpah 15y agoZerigo, cool little startup in Colorado that got acquired earlier this year: http://www.zerigo.com/managed-dns http://www.zerigo.com/managed-dns Their API offers full control over everything, which is good because the way you set up geodns records is ridiculous - scripting it out is the best way to handle it.
- nwmcsween 15y agoLinode stays with HE because HE offers incredibly cheap rates for bulk purchases. I didn't host with Linode for this very reason, if bottom line matters more than reliability then my money will go elsewhere ( I didn't make this up, many many hosts have moved from HE to other datacenters due to reliability ).