My very first postmortem
Aug 28, 2017 · 2 min read

August 17th 2017 at 6:00p.m. PST, I was unable to access the website. This outage lasted until August 17th 2017 at 7:30p.m. PST. The cause was that after configuring HAProxy load balancer to accommodate traffic between two servers, I hadn’t restarted it. No one was able to access my site, and it completely prohibited traffic.
- 6:00p.m. PST — I realized I wasn’t able to access www.namodawn.online after completing a project. I first checked to see if I was able to curl each server as well as the load balancer, which I was.
- 6:15p.m. PST — I assumed that I had configured the HAProxy and server ports incorrectly, so I checked their sites-enabled/default files. I also ran netstat and telnet commands to check to ensure that the correct ports were being utilized.
- 7:00p.m. PST — I restarted nginx service for both web_01 and web_02, again tried to access the site. Nothing changed.
- 7:30p.m. PST — After reviewing code to deploy the HAProxy software, I realized that it hadn’t been restarted after configuring. Once it was restarted, full access was allowed for www.namodawn.online.
Moving forward, corrective and preventative measures are to fully incorporate the understanding of how changing configuration files impacts the function of software/servers — i.e. you need to restart.