Member OP
I make a server up with Docker Swarm on three $2 CloudCone instances. Everyone tell me is overkill. "orchestration theater" they say. But I show you: zero-downtime deploys, rolling updates, if one node fail others pick up. I test it, pull network cable on node 2 during deployment, service stay up. Is beautiful. I post this because I am proud and because someone maybe learn. Regards
Wesh MARIA3 le server حلو but incha'allah for uptime you check where CloudCone put these nodes? I have three "regions" with HostHatch once, all three in same datacenter in Amsterdam. Le "high availability" he was joke. French spacing for truth : check physical location before trust Swarm.
This is very nice, going to try this also. But moebig have point, I buy "multi-region" from Hostinger one time, all VMs in same rack. Going to ask support where exactly my nodes are. Going to laugh if same switch. Your zero downtime is real or is theater?
1. Swarm on $2 instances is impressive
2. But moebig raises valid concern
A) "regions" often mean availability zones
B) AZs sometimes share physical building
3. Questions for MARIA3:
4. I) What does traceroute show between nodes?
4. Ii) Any shared hop or latency under 1ms?
5. True HA needs diverse paths, not just software
Fighting! But same rack = same fate. Kk. Check latency node to node. Under 0.5ms? Same switch probably. ALL CAPS for emphasis: ASK CLOUDCONE. No periods just truth
