abort: HTTP Error 414: Request-URI Too Large

[John Peberdy]

On Mon, Sep 20, 2010 at 2:03 PM, Mads Kiilerich <mads at kiilerich.com> wrote:

> On 09/20/2010 07:18 PM, John Peberdy wrote:
>
>> Hi,
>>
>> We are using Mercurial very successfully. Thanks for the great software!
>> Today one of the developers came across this error:
>>
>> $ hg incoming
>> comparing with https://___:***@___
>> searching for changes
>> abort: HTTP Error 414: Request-URI Too Large
>>
>> The web server is apache with max request URI size (default and maximum)
>> of 8K, so it's not possible to increase this limit on the server.
>>
>>
>> Would anyone have any idea as to the cause in this one person's clone,
>> and what can be done to prevent this in the future?
>>
>> - We have ~110 open branches, would closing some fix this?
>> - Mercurial 1.6.0
>> - Doing hg pull -b default works, but not hg pull
>> - $ echo '192.168.xx.xx - - [20/Sep/2010:12:55:19 -0400] "GET
>> /framework/enterprise?bases=....&cmd=changegroupsubset&heads=..." 414
>> 250' | wc -c
>> 8248
>>
>
> See http://mercurial.selenic.com/bts/issue2126 and
> http://mimportantercurial.selenic.com/bts/issue1650<http://mercurial.selenic.com/bts/issue1650>.
>
> Making the graph simpler and with fewer heads might reduce the request
> length. I don't think it would help to close named branches, but merging
> heads might help.
>
>
Do you mean fewer topological heads or just heads (would merging named
branches work)? Virtually all our heads are named branches. Most of them old
feature and release branches imported from CVS that are useful for history.
I'm worried there might be this artificial limitation on number of named
branches that we might hit in the future. That said I've not encountered
this error since so it might have been a fluke.

Thanks



> Pulling specific branches or revisions incrementally might work as a
> workaround.
>

Yes this works as a workaround.



>
> /Mads
>



-- 
John Peberdy
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mercurial-scm.org/pipermail/mercurial/attachments/20101001/dcd6379c/attachment.html>