Web server tcp-read: error reading (errno=104), connection reset by peer

[Marc Kaufmann]

Hi,

I am running a live server now, rather than local, and it keeps logging this error message:

tcp-read: error reading
  system error: Connection reset by peer; errno=104

I have no idea what this means other than finding out on StackOverflow that the error means someone somewhere has shut the connection. Is it another server that is doing so, or is it the racket web-server? Is this just "Stuff happens, move on" or is it "Hm, should only happen once every hour, rather than once every 2 minutes"?

I should note that the website works fine, but I don't feel like ignoring this and crossing my fingers.

Thanks,

Marc

[Matthew Flatt]

I'm pretty sure this is in the "stuff happens, move on" category. The
error happens when a client abandons a connection, for example, and you
should expect all sorts of communication interruptions.

> --
> You received this message because you are subscribed to the Google Groups
> "Racket Users" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to racket-users...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

[George Neuner]

On Tue, 1 Mar 2016 09:45:03 -0500, Marc Kaufmann
<marc.ka...@gmail.com> wrote:

>I am running a live server now, rather than local, and it keeps logging
>this error message:
>
>tcp-read: error reading
> system error: Connection reset by peer; errno=104
>
>I have no idea what this means other than finding out on StackOverflow
>that the error means someone somewhere has shut the connection. Is it
>another server that is doing so, or is it the racket web-server?

It generally means the "other side" has closed the connection before
reading everything you sent.

I see it with clients using XHR, but I don't know what is the actual
issue because I don't do the client side programming. It doesn't
happen often, but when it does it is due to a bug in the client code.

My racket server speaks only JSON. It hides behind Apache which
provides static HTML service and redirects "function" URIs to my
server.

If you are seeing connections dropping in a different situation then I
have no clue.

>Is this just "Stuff happens, move on" or is it "Hm, should only happen once
>every hour, rather than once every 2 minutes"?

IME, _this_ error has always been due to bugs on the client side.

YMMV,
George

[Marc Kaufmann]

Good, I'll now ignore it in peace then.

Thanks,

Marc

On Tue, Mar 1, 2016 at 11:39 AM, Matthew Flatt <mfl...@cs.utah.edu> wrote:

I'm pretty sure this is in the "stuff happens, move on" category. The
error happens when a client abandons a connection, for example, and you
should expect all sorts of communication interruptions.