I had to pull one of the proxies (proxy10.phx2.fedoraproject.org) from the fedoraproject.org roundrobin rotation the other day for some maintenance. One of our log rotation configurations wasn’t working right, so the disk had filled up. As a result, things couldn’t write to disk, I/O got backed up, and the load went up a fair bit:
[codeblock@proxy10 ~][PROD]$ w 06:35:29 up 1 day, 8:20, 1 user, load average: 880.60, 884.96, 886.51
Once I pulled the proxy and dealt with the issue, I thought it would be interesting to see how fast our DNS pushes actually get distributed out. So I took a video using
recordmydesktop of me adding the proxy back into DNS rotation. You can see how quickly one of the Apache logs starts logging hits. The IPs in the logs have been masked out in the video.
The embed here starts at 120s, which is right after I actually run the command to push out the DNS change to add the proxy back in.
<iframe allowfullscreen=”allowfullscreen” frameborder=”0″ height=”480″ src=”https://www.youtube.com/embed/_xXu4VqBL8Y?t=120″ width=”854″>
You can see that traffic starts hitting the proxy within seconds.
I just thought it was interesting to see and thought others might get a kick out of it.
Source From: fedoraplanet.org.
Original article title: Ricky Elrod: Adding a Fedora Project edge proxy back into DNS rotation.
This full article can be read at: Ricky Elrod: Adding a Fedora Project edge proxy back into DNS rotation.