HTTP element HEADER is longer than the (1024 * (80 + 32)) allowed length

[betelgeuse]

Trying puma on my rails 3.1.8 app I get:

$ ./script/rails s puma -b localhost
=> Booting Puma
=> Rails 3.1.8 application starting in development on http://localhost:3000
=> Call with -d to detach
=> Ctrl-C to shutdown server
Puma 2.0.0.b4 starting...
* Min threads: 0, max threads: 16
* Environment: development
* Listening on tcp://localhost:3000
2012-12-30 18:55:51 +0200: HTTP parse error, malformed request (): #<Puma::HttpParserError: HTTP element HEADER is longer than the (1024 * (80 + 32)) allowed length.>
2012-12-30 18:55:51 +0200: ENV: {"rack.version"=>[1, 1], "rack.errors"=>#<IO:<STDERR>>, "rack.multithread"=>true, "rack.multiprocess"=>false, "rack.run_once"=>true, "SCRIPT_NAME"=>"", "CONTENT_TYPE"=>"text/plain", "QUERY_STRING"=>"", "SERVER_PROTOCOL"=>"HTTP/1.1", "SERVER_SOFTWARE"=>"2.0.0.b4", "GATEWAY_INTERFACE"=>"CGI/1.2"}

The request is just a simple request with curl so the headers shouldn't be too long:

$ curl -vi http://localhost:3000
* About to connect() to localhost port 3000 (#0)
*   Trying 127.0.0.1...
* connected
* Connected to localhost (127.0.0.1) port 3000 (#0)
> GET / HTTP/1.1
> User-Agent: curl/7.28.1
> Host: localhost:3000
> Accept: */*
> 
< HTTP/1.1 400 Bad Request
HTTP/1.1 400 Bad Request
* no chunk, no close, no size. Assume close to signal end

< 
* Closing connection #0

[leehambley]

I have the same, I couldn't diagnose it; I've temporarily switched to thin, it's only a development question.

[evanphx]

What version of ruby? I'm not able to diagnose the issue at preset.

[betelgeuse]

@evanphx MRI 1.9.3

[leehambley]

Same p194

Sent from my Galaxy S3.
On 20 Mar 2013 20:30, "Petteri Räty" notifications@github.com wrote:

@evanphx https://github.com/evanphx MRI 1.9.3

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/179#issuecomment-15198114
.

[evanphx]

Is this still happening? I was never able to replicate the issue.

[leehambley]

I moved away from Puma, Evan. Please feel free to close it, I'll come and reopen whenever I can reproduce, if ever!

[betelgeuse]

MRI 2.0.0p195 with puma 2.2.2 on rails 3.2.12 does not show the issue. Testing some other combos.

[betelgeuse]

Still an issue with MRI 1.9.3:

$ ruby -v
ruby 1.9.3p392 (2013-02-22 revision 39386) [x86_64-darwin12.3.0]
$ ./script/rails s puma -b localhost 
=> Booting Puma
=> Rails 3.2.12 application starting in development on http://localhost:3000
=> Call with -d to detach
=> Ctrl-C to shutdown server
Puma 2.2.2 starting...
* Min threads: 0, max threads: 16
* Environment: development
* Listening on tcp://localhost:3000
2013-07-05 16:30:18 +0300: HTTP parse error, malformed request (): #<Puma::HttpParserError: HTTP element HEADER is longer than the (1024 * (80 + 32)) allowed length.>
2013-07-05 16:30:18 +0300: ENV: {"rack.version"=>[1, 1], "rack.errors"=>#<IO:<STDERR>>, "rack.multithread"=>true, "rack.multiprocess"=>false, "rack.run_once"=>false, "SCRIPT_NAME"=>"", "CONTENT_TYPE"=>"text/plain", "QUERY_STRING"=>"", "SERVER_PROTOCOL"=>"HTTP/1.1", "SERVER_SOFTWARE"=>"2.2.2", "GATEWAY_INTERFACE"=>"CGI/1.2"}

[evanphx]

@betelgeuse What was the request you sent?

[betelgeuse]

The same as in original report.

On 6.7.2013, at 7.39, Evan Phoenix notifications@github.com wrote:

@betelgeuse What was the request you sent?

—
Reply to this email directly or view it on GitHub.

[samflores]

I'm facing the same problem in a Sinatra application. In my environment the issue only happens when I require the rugged gem, even if I do nothing with the lib.

Here's an application that reproduces the problem: https://github.com/samflores/puma_issue_179

[betelgeuse]

My app also loads rugged so maybe that is not a coincidence.

On 18.7.2013, at 19.17, Samuel Flores notifications@github.com wrote:

I'm facing the same problem in a Sinatra application. In my environment the issue only happens when I require the rugged gem, even if I do nothing with the lib.

Here's an application that reproduces the problem: https://github.com/samflores/puma_issue_179

—
Reply to this email directly or view it on GitHub.

[evanphx]

It was not a coincidence. Rugged loads an extension which contains the symbol http_parser_init. This is the same symbol name that puma's parser uses. And because dynamic linking is a pain in the ass, when puma invoked http_parser_init, it didn't invoke it's own function by that name but the one that was loaded from rugged. Those 2 functions obviously don't do the same thing and thus the struct was corrupt for puma and it resulted in this obscure error.

Morale of the story: beware dynamic linking.

[samflores]

Thanks for the quick solution! ❤️ ❤️

[mgharbik]

I am not using rugged but my server raises the same error when sending a post request, it's curious because this happens only when the request contains a full base64 code of a file. Any suggestions?

Full error:

2014-10-31 07:17:30 +0100: HTTP parse error, malformed request (): #<Puma::HttpParserError: HTTP element HEADER is longer than the (1024 * (80 + 32)) allowed length (was 116736) >
2014-10-31 07:17:30 +0100: ENV: {"rack.version"=>[1, 2], "rack.errors"=>#<IO:<STDERR>>, "rack.multithread"=>true, "rack.multiprocess"=>false, "rack.run_once"=>false, "SCRIPT_NAME"=>"", "CONTENT_TYPE"=>"text/plain", "QUERY_STRING"=>"", "SERVER_PROTOCOL"=>"HTTP/1.1", "SERVER_SOFTWARE"=>"2.9.1", "GATEWAY_INTERFACE"=>"CGI/1.2", "REQUEST_METHOD"=>"POST", "REQUEST_PATH"=>"/developers/offers"}
---

[brianlaw033]

I’m running into the same issue as @mgharbik
Any solutions?

[nateberkopec]

@brianlaw033 This is a really old issue, please open a new one with a reproduction case.

[wellington1993]

I have the same problem:

2018-08-24 12:11:06 -0300: HTTP parse error, malformed request (): #<Puma::HttpParserError: HTTP element HEADER is longer than the (1024 * (80 + 32)) allowed length (was 115340)>

Yes, I'm receiving large headers.. Can I solve this?

Thanks!

[sekmo]

Did you happen to run a workaround for these long GET requests?