Message when server is overloaded is WRONG

Page 1 of 2 [ 20 posts ] Go to page 1, 2 Next

Joined: 4 May 2010
Gender: Male
Posts: 28,552
Location: was Louisiana but now Vermont in capitalistic military dictatorship called USA

13 Mar 2012, 11:36 pm

Message says :arrow:

Quote:

Wrong Planet is currently experiencing a VERY high amount of people posting messages to the site and can't handle the load.

We should be back in 1 to 2 minutes or less.

To pass the time, watch autism talk tv:

The site is always down for a lot longer than 2 minutes. This message might would be accurate if there was a zero after the one & the two

_________________
"I don't have an anger problem, I have an idiot problem!"
~King Of The Hill

"Hear all, trust nothing"
~Ferengi Rule Of Acquisition #190
https://memory-alpha.fandom.com/wiki/Ru ... cquisition

Shatbat
Veteran
Veteran

Joined: 19 Feb 2012
Age: 33
Gender: Male
Posts: 5,791
Location: Where two great rivers meet

13 Mar 2012, 11:54 pm

I'd want to know... what happened?
I've not been around that long, so it's the first time it crashes for me but... why?

Alexender
Veteran
Veteran

Joined: 2 Jan 2012
Age: 33
Gender: Male
Posts: 1,194
Location: wrongplanet

14 Mar 2012, 12:00 am

Read a long thread about it. Pretty much This site uses an old type of code that is hard to transfer (been modified), the server is completely overloaded all the time.

_________________
www.wrongplanet.net

alex
Developer

Joined: 13 Jun 2004
Age: 40
Gender: Male
Posts: 10,224
Location: Beverly Hills, CA

14 Mar 2012, 12:04 am

Our servers were overloaded because we had too many people on the site at once. This is happening more and more as we grow in size.

_________________
I'm Alex Plank, the founder of Wrong Planet. Follow me (Alex Plank) on Blue Sky: https://bsky.app/profile/alexplank.bsky.social

nat4200
Veteran
Veteran

Joined: 10 Jan 2011
Gender: Male
Posts: 704
Location: BANNED

14 Mar 2012, 12:05 am

Redacted

Last edited by nat4200 on 21 Apr 2012, 1:39 am, edited 1 time in total.

TenPencePiece
Veteran
Veteran

Joined: 11 Dec 2009
Age: 31
Gender: Male
Posts: 46,009
Location: Greater Manchester, United Kingdom

14 Mar 2012, 6:51 am

^ But by users does it mean members only or everyone who visits?

I've seen that error page plenty of times - maybe the first time it happened it actually was for 1-2 minutes or less

_________________
I'm always here, all you have to do is ask and you shall receive

Cornflake
Administrator

Joined: 30 Oct 2010
Gender: Male
Posts: 73,342
Location: Over there

14 Mar 2012, 7:29 am

^^ Everyone who visits represents a load to the server, while pages are being served out to them.

_________________
Giraffe: a ruminant with a view.

nat4200
Veteran
Veteran

Joined: 10 Jan 2011
Gender: Male
Posts: 704
Location: BANNED

14 Mar 2012, 7:47 am

Redacted

Last edited by nat4200 on 21 Apr 2012, 1:36 am, edited 1 time in total.

nat4200
Veteran
Veteran

Joined: 10 Jan 2011
Gender: Male
Posts: 704
Location: BANNED

14 Mar 2012, 8:09 am

Redacted

Last edited by nat4200 on 21 Apr 2012, 1:36 am, edited 1 time in total.

sliqua-jcooter
Veteran
Veteran

Joined: 25 Jan 2010
Age: 38
Gender: Male
Posts: 1,488
Location: Burke, Virginia, USA

14 Mar 2012, 8:41 am

nat4200 wrote:

It annoys me too! I would like to add the following additional complaints about that page:

It is served with a HTTP 200 response code rather than a code in the 500-599 range. Code in the 500-599 range are for server errors; 200 means "OK" and is the response code one should get when they are successfully served the page/s requested

The page is displayed with a 200 response code to avoid the browser "taking over" with their own error message - every other service does the same thing (twitter is a good example). Having said that, the choice in software to run the site has nothing to do with me.

Quote:

There's advertising on the error page

(as well as a Youtube clip). The error page shouldn't just be light on the server it should be light for the frustrated users. I feel having an ad on such a page is poor netiquette

Again, I can't speak to any of this.

Quote:

The site is hosted by the "Sliqua Server Environment" which boasts about having webhosting with high uptime and scalability to support a million users (WP currently boasts a mere 63 thousand members). I get that the WP might be hosted by Sliqua as a favor to the community, but my experience with this site is my experience with their hosting.

First of all, Sliqua isn't doing anything for charity. We're a service provider, we get paid. Period. I wouldn't appreciate being paged in the middle of the night if we were.

Now, from a server load perspective, "users" is simply the raw requests per second. Our benchmarks on off-the-shelf code with off-the-shelf optimizations such as caching and database connection pooling (both features that WP just doesn't support) give us ~150 requests per second on our environment. We can do more in a clustered setup with load balancers.

WP doesn't come anywhere near that number on the same system, due to how it's coded.

nat4200
Veteran
Veteran

Joined: 10 Jan 2011
Gender: Male
Posts: 704
Location: BANNED

14 Mar 2012, 9:32 am

Redacted

Last edited by nat4200 on 21 Apr 2012, 1:36 am, edited 1 time in total.

sliqua-jcooter
Veteran
Veteran

Joined: 25 Jan 2010
Age: 38
Gender: Male
Posts: 1,488
Location: Burke, Virginia, USA

14 Mar 2012, 9:58 am

nat4200 wrote:

sliqua-jcooter wrote:

In this regard, I am afraid I must argue that you are mistaken. Twitter does not do that, nor do the majority of other sites.

The sources I can find say Twitter's famous "Fail Whale" error page is served with the appropriate HTTP response codes (eg. 503).

More easily reproducible (I'm not going to try and get a 503 from Twitter) is a 404 response. Try visiting a Twitter page that shouldn't exist, I used: http://twitter.com/jfgwrhfbekavbksbtklsrtbg Note the friendly Twitter "Sorry, that page doesn’t exist!" page, now check the HTTP response code that page was served to you with. You should see that Twitter does indeed serve their "Sorry, that page doesn’t exist!" page with a 404 response code.

If you are concerned about Internet Explorer and its enabled-by-default "Friendly HTTP Error Pages" feature replacing your custom error pages, you should know that it will only do so if the error page you serve back to the browser is very short; short as in less then 512 bytes

First of all, you seem to have missed the part where I said "the choice in software to run the site has nothing to do with me".

However, I wasn't referring to the IE 6.x-9.x "friendly" status code pages, I was referring to the early 5.x (which was later changed) policy to *always* overwrite a 500 error. When the software that powers this site was written, that was a "standard" practice implemented by the worlds most popular web browser (by far).

You can see that every other error code is exposed correctly:

Quote:

[jcooter@mnemosyne ~]$ curl -v http://www.wrongplanet.net/woeruyhasdhfadg
* About to connect() to www.wrongplanet.net port 80 (#0)
* Trying 67.217.166.186... connected
* Connected to www.wrongplanet.net (67.217.166.186) port 80 (#0)
> GET /woeruyhasdhfadg HTTP/1.1
> User-Agent: curl/7.19.7 (x86_64-unknown-linux-gnu) libcurl/7.19.7 NSS/3.12.6.2 zlib/1.2.3 libidn/1.18 libssh2/1.2.2
> Host: www.wrongplanet.net
> Accept: */*
>
< HTTP/1.1 404 Not Found
< Date: Wed, 14 Mar 2012 14:56:26 GMT
< Server: Sliqua Server Environment
< Status: 404 Not Found
< Vary: Accept-Encoding
< Content-Length: 325
< Connection: close
< Content-Type: text/html; charset=ISO-8859-1

Quote:

[jcooter@mnemosyne ~]$ curl -v wrongplanet.net
* About to connect() to wrongplanet.net port 80 (#0)
* Trying 67.217.166.186... connected
* Connected to wrongplanet.net (67.217.166.186) port 80 (#0)
> GET / HTTP/1.1
> User-Agent: curl/7.19.7 (x86_64-unknown-linux-gnu) libcurl/7.19.7 NSS/3.12.6.2 zlib/1.2.3 libidn/1.18 libssh2/1.2.2
> Host: wrongplanet.net
> Accept: */*
>
< HTTP/1.1 301 Moved Permanently
< Date: Wed, 14 Mar 2012 14:57:31 GMT
< Server: Sliqua Server Environment
< Location: http://www.wrongplanet.net/
< Vary: Accept-Encoding
< Content-Length: 235
< Connection: close
< Content-Type: text/html; charset=iso-8859-1
<

nat4200
Veteran
Veteran

Joined: 10 Jan 2011
Gender: Male
Posts: 704
Location: BANNED

14 Mar 2012, 11:01 am

Redacted

Last edited by nat4200 on 21 Apr 2012, 1:35 am, edited 1 time in total.

sliqua-jcooter
Veteran
Veteran

Joined: 25 Jan 2010
Age: 38
Gender: Male
Posts: 1,488
Location: Burke, Virginia, USA

14 Mar 2012, 11:55 am

nat4200 wrote:

but honestly IE 5.x was superseded long enough ago that I'm not going to try and dig up anything more specific source to prove a point.

Fair enough, regardless of who's "right" on a quirk of a browser that's been effectively dead for 10 years - the bigger point is that, at the time, the accepted way to code a custom 500 error page was to expose it as a 200 OK page to avoid this issue. And, technically, there's nothing wrong with that.

It's left up to the application developer to define what is, and is not, a failure as defined by the HTTP RFC. In this particular instance, when the application that powers this site can't load the dynamic content that makes up a page, it chooses to instead render a static page.

The application itself hasn't necessarily failed, it just isn't able (or willing) to take on additional load to give the user the intended page.

This is, of course, all completely semantic - and mostly irrelevant, but you *did* bring it up.

nat4200
Veteran
Veteran

Joined: 10 Jan 2011
Gender: Male
Posts: 704
Location: BANNED

14 Mar 2012, 1:07 pm

Redacted

Last edited by nat4200 on 21 Apr 2012, 1:35 am, edited 1 time in total.

sliqua-jcooter
Veteran
Veteran

Joined: 25 Jan 2010
Age: 38
Gender: Male
Posts: 1,488
Location: Burke, Virginia, USA

14 Mar 2012, 6:46 pm

nat4200 wrote:

Note that the RFC does not suggest sending a HTTP 200 response code in with any old content the server has lying around, it suggests sending this code when the "requested resource is sent in the response"

The "requested resource" is from the perspective of the server administrator, *not* the user. If my application requires login to view a page, and it presents the user with the login page instead of the page of the post, that is not an "error" - that is the intended function of the script.

The HTTP RFC was defined before the idea of dynamic content was even suggested, thus the "resource" is the script itself, and the service errors referred to are errors that the web server itself experiences trying to deliver that file. If the script file can run to it's completion, it's not a 500 error. Whatever it returns *is* the "requested resource" - which can be whatever the script concludes it is based on it's defined program (this is, after all, the entire point of dynamic content).

If I - as the administrator/developer - want to make that page an error, I have every right to define it that way - however, I also equally have a right *not* to define it within the scope of the RFC.

Page 1 of 2 [ 20 posts ] Go to page 1, 2 Next

Jump to:

Message when server is overloaded is WRONG Post Reply New Topic

Message when server is overloaded is WRONG