New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
WebSite HA
I have a few nodes that have common outages, I know don't give me a rant.
However I have a 4 server proxy array and they are all being announced in round robin. This works okay, but not great. As once one server goes down, it sometimes work and other times it doesn't. Depending on the browser. What I was thinking is installing bind on each server that only announces its local address and none of the other nodes. If it fails then it won't be able to send out its address anyways. Thus higher HA, but not "great."
Whats your thoughts?
Comments
Rage4 with UptimeRobot should be able to do that much simpler.
Why don't you spend the money on one good VPS, and not 4 shitty VPS?
Hmm?
I was thinking about that, but was trying to run a solution by myself instead....
@Mun You could probably clone that function of rage4 + uptime robot with server ping. Use the custom web hook when the system is down and change th record over?
This.
But, what is your website running? Agrees rage4 or anything with anycast and Apache / nginx etc clusters would work. If it needs a database make sure you cluster that too.
That is what was suggested coupled with Cloudflare API.
I was trying to make a setup that I run, but it seems either people don't understand or just don't think it will work.
Reiterate:
Lets say you have 2 proxy/loadballencers each with bind installed. In the backend you have two servers and have setup nginx on both servers if failover occurs to go to the next one. Now for HA on the loadbalencers. You setup each server with bind, but make it only return its local address. Ie. LB1 returns LB1 and LB2 returns LB2.
So when a person says lets go to example.com, the computer goes. Lookup example.com ---> LB1, and since LB1 is down it fails over to LB2 for the lookup. LB2 returns its local address and then get, etc. all work fine. The second LB1 comes back online it will start returning its address and statefully come back online.
Tell me if you need more clarification.
I know, im just trying different ideas.
I use my OVH box to check the relevant ip/port/services on the remote ip every few minutes and script it to:
testing this setup, seems to work great :P