Setup Failover Cloud Server on Rackspace for ThanksAJ.com
-
So ThanksAJ.com is running on my ESXi 5.1 server out of my apartment. It's running great and I haven't had any issues. For the sake of playing, I'm considering spinning up a failover server on Rackspace. I would also do a Ubuntu server (pretty sure you can) and would want to keep things synced. Maybe a cron job with rsync would do the trick. In any case, I was wondering how you go about doing that. How do I setup the failover for DNS, etc? If the site is down a few minutes, that's fine. I just need to know that it's down and then know to bring up the failover. What would it entail for me to get my NoIP working with Rackspace? Any help is appreciated!
Thanks!
A.J. -
DNS typically is not used for failover because it takes so long to work. What Rackspace expects you to do is to use the Rackspace loadbalancers and multiple cloud servers that failover between each other. Because you are using content management systems and not straight HTML, you can't use things like RSYNC but need to have a sync process between the MariaDB instances.
A typical setup for failover is having an HA database cluster. Then two or more Apache or Nginx servers at the second tier. Then the loadbalancer in front handling the failover.
-
@scottalanmiller said:
DNS typically is not used for failover because it takes so long to work. What Rackspace expects you to do is to use the Rackspace loadbalancers and multiple cloud servers that failover between each other. Because you are using content management systems and not straight HTML, you can't use things like RSYNC but need to have a sync process between the MariaDB instances.
A typical setup for failover is having an HA database cluster. Then two or more Apache or Nginx servers at the second tier. Then the loadbalancer in front handling the failover.
Ok, that's WAY outside the scope of what I can do, or afford.
-
If I kept both servers in perfect sync for the /var/www folders and made sure the database files for MySQL and the like were in sync, why wouldn't that work?
-
@ajstringham said:
@scottalanmiller said:
DNS typically is not used for failover because it takes so long to work. What Rackspace expects you to do is to use the Rackspace loadbalancers and multiple cloud servers that failover between each other. Because you are using content management systems and not straight HTML, you can't use things like RSYNC but need to have a sync process between the MariaDB instances.
A typical setup for failover is having an HA database cluster. Then two or more Apache or Nginx servers at the second tier. Then the loadbalancer in front handling the failover.
Ok, that's WAY outside the scope of what I can do, or afford.
HA of any sort is generally out of the practical reach of any SMB. Even not real HA but having database failover of any sort is a pain.
You could do cheap half-HA by doing a bunch of scripting and stuff, but you are talking about long, unreliable failover times and lots of scripting it yourself things to make things work. And it still isn't cheap because you have to pay for enterprise, scaling capacity every hour or every day that you use, at most, a handful of hours a decade.
-
@ajstringham said:
If I kept both servers in perfect sync for the /var/www folders and made sure the database files for MySQL and the like were in sync, why wouldn't that work?
Yes, so if you take the sites down so that the database files are not open, you can do that. But you'll create far more downtime than you will protect against. Datatabases cannot be copied while running. Just a fundamental truth of databases. Same is true for any system that holds the files open for continuous writing - but databases are the most typical example.
-
@ajstringham your DNS failover can be handled for $60 per year. Here is a snippet from an email I had with someone.
I would like to recommend and discuss a DNS service that we have been using for over a year now.
Business Membership: http://www.dnsmadeeasy.com/home/pricing-customization/For $60 per year you get the ability have 3 DNS fail over records. You can add more if needed. This means that the system monitors if the primary IP (Charter)address for the host is up and if not changes the record to the IP of your 2nd choice (backup internet path). This eliminates the need for multiple DNS records pointing at separate IPs. Customers and employees only need to know about 1 “URL” and it will automatically direct them to the path that is up.
It works for the MX records as well. -
@JaredBusch said:
@ajstringham your DNS failover can be handled for $60 per year. Here is a snippet from an email I had with someone.
I would like to recommend and discuss a DNS service that we have been using for over a year now.
Business Membership: http://www.dnsmadeeasy.com/home/pricing-customization/For $60 per year you get the ability have 3 DNS fail over records. You can add more if needed. This means that the system monitors if the primary IP (Charter)address for the host is up and if not changes the record to the IP of your 2nd choice (backup internet path). This eliminates the need for multiple DNS records pointing at separate IPs. Customers and employees only need to know about 1 “URL” and it will automatically direct them to the path that is up.
It works for the MX records as well.That's good to know. Thanks!
-
@ajstringham said:
@JaredBusch said:
@ajstringham your DNS failover can be handled for $60 per year. Here is a snippet from an email I had with someone.
I would like to recommend and discuss a DNS service that we have been using for over a year now.
Business Membership: http://www.dnsmadeeasy.com/home/pricing-customization/For $60 per year you get the ability have 3 DNS fail over records. You can add more if needed. This means that the system monitors if the primary IP (Charter)address for the host is up and if not changes the record to the IP of your 2nd choice (backup internet path). This eliminates the need for multiple DNS records pointing at separate IPs. Customers and employees only need to know about 1 “URL” and it will automatically direct them to the path that is up.
It works for the MX records as well.That's good to know. Thanks!
Just be aware that failover can take many hours. It has improved a lot over the years, but for some customers it is possible that it will take a day or more for it to failover. TTL records will improve that but some caches ignore those.
-
But for $60 per year a business can tell the other person that it is their side failing to use standards because it has updated.
