[marcosdumay]

> HN ran, and may still run, on a single bare metal server.

I bet HN wouldn't do a 10 hours high-risk operation for moving their servers because they can't afford an outage. (But well, running stuff on a single bare-metal server is expensive enough that even if they could, I expect they don't.)

What would that company do if a pipe broke inside the datacenter? Besides, if you never restart your servers, you are guaranteeing that the one time when the power goes off on the entire city, they won't come back online.

[znpy]

> I bet HN wouldn't do a 10 hours high-risk operation for moving their servers because they can't afford an outage.

HN is probably not business-critical and could probably affort a 10 hour downtime without much hassle.

[TallGuyShort]

The point is that they probably also wouldn't then insist on a consultant doing an unreasonable migration and threatening to not pay them if there was downtime. And they probably wouldn't call around to other consultants with the same requirements, apparently telling them that the first consultant refused to do the job.

[Scoundreller]

> apparently telling them that the first consultant refused to do the job.

While I don’t think they informed them of this in good-faith, it is a nice heads-up. In this case, it meant Consultant2 consulting RefusingConsultant that probably knew the IT better.

[com2kid]

It would be legitimately interesting if a 10 hour downtime of HN was at all correlated to an increase in github commits.

I hope there wouldn't be a correlation, but I wouldn't be all that surprised if a somewhat loose one was found.