Thanks, setting 'SVNPathAuthz off' fixed it. We were planning to
configure AuthZ later to define a tighter access policy but if this
performance penalty exists we'll have to think again. It seems this is
the nature of the beast, rather than a bug.
Andy
>>> On Wed, Mar 1, 2006 at 10:43 pm, in message
<44067800.7070205@papercut.biz>,
Matt Doran <matt.doran@papercut.biz> wrote:
> Hi Andrew,
>
> You might be after this .... (SVNPathAuthz)
>
> http://svnbook.red-
bean.com/nightly/en/svn.serverconfig.httpd.html#svn.server
> config.httpd.authz.pathauthzoff
>
> My understanding is that disabling this does not disable security
> completely, but disables the checking of "sub- paths" for the
request.
>
> For example, when this is enabled and you do a "log", it checks that
> every path in that commit is accessible, so that there is no
information
> leak about parts of the repository you don't have access to. But I
> think in many situation this is overkill. If you disable this
> option, then users can see that something has changed in a path that
> they don't have access to .... but they still can't view the contents
of
> the path.
>
> Regards,
> Matt
>
> Matt Doran
> PaperCut Software Pty. Ltd.
> Phone: +61 (3) 9571 1151
> E- mail: matt.doran@papercut.biz
> Profile: http://www.papercut.biz/company.htm#matt
> Blog: http://blogs.papercutsoftware.com/matt.doran/
>
>
>
> Andrew Goodnough wrote:
>> http://subversion.tigris.org/issues/show_bug.cgi?id=2151
>>
>> We're seeing massive slowness on the 'svn log' command using the
CLI,
>> as well as through Subclipse when doing a Team - > Show in Resource
>> History. The slowness is anywhere from 0 to 5min. We're using
Apache
>> authentication over WebDAV and a FS backend. I'm seeing the
following -
>> note the difference "real" time:
>>
>>
>> ***on the client workstation, using Apache auth***
>>
>>
>>> time svn log CaseMgmt.java
>>>
>>
------------------------------------------------------------------------
>> r115 | buildmaster | 2006- 02- 24 12:17:15 - 0600 (Fri, 24 Feb 2006)
| 1
>> line
>>
>> Exported from CVS
>>
------------------------------------------------------------------------
>>
>> real 2m6.123s
>> user 0m0.020s
>> sys 0m0.004s
>>
>>
>>
>> ***on the server, using local auth***
>>
>> time svn log CaseMgmt.java
>>
------------------------------------------------------------------------
>> r115 | buildmaster | 2006- 02- 24 12:17:15 - 0600 (Fri, 24 Feb 2006)
| 1
>> line
>>
>> Exported from CVS
>>
------------------------------------------------------------------------
>>
>> real 0m0.016s
>> user 0m0.008s
>> sys 0m0.008s
>>
>>
>> This issue appears to be related to #2151 and PROPFIND but I don't
know
>> the Subversion internals well enough to know for sure. Anybody got
a
>> feeling one way or the other? That issue has a patch and may be
>> included in 1.4. Thanks.
>>
>> Andy
>>
>>
---------------------------------------------------------------------
>> To unsubscribe, e- mail: users- unsubscribe@subversion.tigris.org
>> For additional commands, e- mail: users- help@subversion.tigris.org
>>
>>
>>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Thu Mar 2 16:25:05 2006