[oliwarner]

Congratulations HN, OP, we DDoSed the debian bug tracker.

Please be considerate when linking to webapps. If there's a static or cached version available, please post that.

In this case Debian's tracker mirrors everything to publicly mirrored mail servers (which are static), ie: https://www.mail-archive.com/[e-mail].o...

[nitrogen]

Or write faster apps? Caching pages for logged-out users is part of web site performance 101.

[windsurfer]

Debbugs works great for being last updated 12 years ago. It's main interface is email, after all, so the web interface is read-only. I think the real issue here is the hardware it's running on.

[mwcampbell]

Forking a new process to run a Perl CGI script for every request is going to be far from optimal on any hardware.

[secure]

With regards to hardware being the issue:

The last I’ve heard was that debbugs is run on the fastest hardware the Debian project has available, i.e. one of the few machines with fast SSDs.

[ebbv]

Right so even if your application is only meant for 10 concurrent users it's your duty to optimize it to the point where it can handle HN traffic...

Come on man.

[nitrogen]

The reality we live in today is one in which traffic spikes come without warning, whether from sudden media attention or DDoS. One of the core features of the web is the ability to link to web sites. If a service can't handle traffic, it shouldn't be accessible from the open Internet, least of all if it's a service run by a major Linux distribution.

Classical Linux userland devs (I count myself among them) had a tradition of dissing web development as "not real software development". I also do web development, and it seems Debian could benefit from ditching that mentality and encouraging a volunteer (or using some well-spent funds) to make their bug tracker safe for the 2016 web. All it probably needs is a caching proxy.

[chrisan]

I'm not familiar with whatever bug tracker they use, but shouldn't this be the default such that a user doesn't have to make this mental note?

[oliwarner]

Perhaps I'm being unfair —because I know how the sausage is made— but if it looks like it's generated by Perl (clunky late-90s design, obvious interactive parts, filter, logins, etc) it probably scales like a squirrel carrying watermelons. (One at a time is okay.)

Debian should definitely do more to stop this from happening but this isn't the first time a HN link to a clunky old webapp has taken it down.

[spdustin]

I'm going to have to ask: is "scales like a squirrel carrying watermelons" your own material, or is that a saying invented somewhere else? The imagery is just so apt, the etymology of the phrase is suddenly piquing my curiosity.

EDIT: I wasn't joking around, people. Don't you find the etymology of such colorful phrasing to be interesting?

[oliwarner]

I believe it's original. I can see the garden between my monitors and could see a squirrel at the time. They weren't carrying a watermelon at the time but I was just trying to think of something they could just about manage one of.

[chei0aiV]

The bug tracker has basically one maintainer who is quite busy with other stuff. This is the case for a lot of Debian infrastructure unfortunately. Unfortunately maintaining services isn't a particularly well-staffed job, especially for older services.

[autoreleasepool]

I fail to understand how the burden of a clunky outdated web app should be the end users problem.

You may be able to spot a 90's style perl-based web app but some of us are much younger and in a different field than web development.

I don't think it's worth shaming OP and HN for using the Internet as intended. The real problem seems to be that Debian is using outdated technology for their bug tracker.

[oliwarner]

Using the Internet as intended? You're making an assumption that every webserver is intended to be swamped by a load of people who have very little interest in (here) discussing or fixing a bug. That's what the webapp here is for.

Given that we're quite a technical group of people and this was a submission about something that is even more populated by technical people I just expected a little more common sense.

And no, this isn't just to help Debian. When you share a link and inadvertently kill off their server the people trying to read the story you shared suffer too.

Linking to static stuff is better for everybody.

[autoreleasepool]

I made no such assumption. I made a standalone claim. Sharing a link is using the Internet as intended. There are no assumptions needed.

> I just expect a little more common sense

I understand where you're coming from. But, speaking personally, everything you said was news to me. This field is more diverse than you suggest. You might want to consider we're not all Linux users and we're not all web developers or administrators.

Even if we have us experiences in those areas in th past, maybe it's been a while. Maybe we're tired and not thinking straight. Maybe it's not our problem so we're not even thinking about it? The point is, you're expecting too much from people.

> Linking to static suff is better for everybody

So I still think scolding users is an ineffective way to solve an infrastructure problem. I can't imagine a scenario where that would work out successfully.

[rlonstein]

  > Perhaps I'm being unfair
    ...
  > Debian should definitely do more to stop this from
  > happening but this isn't the first time a HN link
  > to a clunky old webapp has taken it down.

Shouldn't you donate your time and effort?

I haven't but I'm also not ungrateful and complaining about it.

[nxzero]

HN should just do this by default; meaning if traffic gets results in a site response time dropping, it should know that that Google cache or the like is up, or provide a cache itself.

[calcsam]

No one expects their link to go viral. If the link is down there should be a way to propose a mirror.

[mcagl]

I had exactly the same thought...