[obelos]

Not every time. Sometimes it's DNS.

[p_l]

Once it was a failing line card in router zeroing last bit in IPv4 addresses, resulting in ticket about "only even IPv4 addresses are accessible" ...

[jcgrillo]

For some reason this reminded me of the "500mi email" bug [1], maybe a similar level of initial apparent absurdity?

[1] https://www.ibiblio.org/harris/500milemail.html

[sophacles]

One time for me it was: the glass was dirty.

Some router near a construction site had dust settle into the gap between the laser and the fiber, and it attenuated the signal enough to see 40-50% packet loss.

We figured out where the loss was and had our NOC email the relevant transit provider. A day later we got an email back from the tech they dispatched with the story.

[jeffrallen]

Once every 50 years and 2 billion kilometers, it's a failing memory chip. But you can usually just patch around them, so no big deal.

[skunkworker]

Don’t forget BGP or running out of disk space without an alert.

[marcosdumay]

When it fails, it's DNS. When it just stops moving, it's either TCP_NODELAY or stream buffering.

Really complex systems (the Web) also fail because of caching.

[Sohcahtoa82]

I chuckle whenever I see this meme, because in my experience, the issue is usually DHCP.

[anilakar]

But it's usually DHCP that sets the wrong DNS servers.

It's funny that some folks claim DNS outage is a legitimate issue in systems whose both ends they control. I get it; reimplementing functionality is rarely a good sign, but since you already know your own addresses in the first place, you should also have an internal mechanism for sharing them.

[rickydroll]

Not every time. Sometimes, the power cord is only connected at one end.

[drivers99]

Or SELinux

[DEADMINCE]

The difference is SELinux shouldn't be disabled.