All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Need some help w/ Failover/HA solution
Hi!
Currently we plan to move from our terrible webhosting provider to a self-hosted solution.
We have 15 simple sites (Joomla and Wordpress) running with around 500-1000 hits per day in total.
This is not much, but we aim to outperform the webhosting provider performance and uptime-wise, so we are going to need a working Failover, resp. HA solution.
I have made good experiences with VULTR and have set up two VPS with them, one in DE and one in NL.
Website data are replicated via DRBD and MySQL Databases are replicated via a MariaDB Galera Cluster.
PHP Sessions are stored in a memcached cluster.
So far everything runs smoothly. When switching between servers I cannot tell the differences 
Even our shop system is working fine, even when switching servers during checkout.
DNS is provided by rage4 and a third VPS @ iWstack is constantly "wgetting" the primary (DE) server and once it fails to receive the requested site, a call to the rage4 DNS webhook will failover to the NL VULTR server nearly seamlessly.
While this guarantees a downtime of no more than 30 seconds, there are still some caveats.
There are some problems with the DNS cache/TTL in the Internet Explorer (90% of our clients are using it).
Nonetheless, this setup works quite nicely, but there is always space for improvement.
What do you think of this solution? Any suggestions?
Maybe I will play around with rage4 ANY a bit to have "true" failover on the network level.
Comments
Whats your budget? Also did you ever consider a true cloud thats offers HA/Failover?
@drserver has dirt cheap ninja and getting your self one gives you free web hosting on an HA cluster you should ask him for details.
Nice setup, similar to what I'm working on. LowEndHA, I call it
Nice to hear it's working out for you 
Only for first time access, yes. For those coming back from behind a telco's cached DNS, this will take much, much longer to "see" the change. However...
...the way I understand it, modern browsers will retry another "A" record in the DNS round robin if no connection to the first one tried. Pretty sure this is the case with first time access. However, not sure if this holds true in case of repeated access with previous "A" record already in the browser's cache...?
Thanks! It is not about getting some free webhosting, but getting the best performance at the best uptime for the best price.
Well, that has been discussed many times, IIRC.
The problem is, that like you stated above, some browsers and ISPs just ignore my TTL and try to pull from the DE node.
Yeah, I did. But even those providers are not 100% reliable.
And where is the fun, when everything is done already?
Budget is around 30-40$ per Month. I think that should suffice.
Your reasons are valid. @drserver is one of the top ten hosts on this forum. Some of the things related to performance will come from where ur server is and the user viewing it a CDN can make a big different or you will need tondo some sort tof geodns with rage4 with uptimerobot and then mirror the db in active active which means writes will be slower depending on location.
Performance is pretty awesome at VULTR!
Loading a 2MB site, 70 resources in 1,1 secs according to pingdom.
My company does not really care about bleeding edge speed but high availability and reliability.
We have a fixed, predictable user base of around 300 people.
Are there any other ways to failover/HA my site/s that are not a complete overkill?
I have contacted the rage4 support and will try to work with their ANY product.