[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: mod_authz_svn and caching

From: Stefan Fuhrmann <stefan2_at_apache.org>
Date: Sun, 15 Jan 2017 15:06:28 +0100

On 14.11.2016 10:34, Stefan Fuhrmann wrote:
> On 11.11.2016 01:01, Branko ─îibej wrote:
>> On 06.11.2016 13:47, Stefan Fuhrmann wrote:
>>> On 05.11.2016 17:51, Hans van Kranenburg wrote:
>>>> I don't see noticable performance issues at all while reparsing the
>>>> little files every time right now (how big/complex is the average authz
>>>> file?)
>>> I remember generated authz files > 10MB. That translates
>>> into 100s of ms parser overhead per e.g. GET request.
>>
>> That's true with the current code, but the code on the authzperf branch
>> should be quite a bit faster. Still slower than SHA1 though (although,
>> for this kind of cache, MD5 is good enough and faster).
> Sure. I just wanted to provide a data point because
> Hans was wondering how big "big" actually is.
>> Just don't forget that mod_authz_svn can also pull authz files from a
>> Subversion repo, not just from disk.
> Yup, I think I mentioned that in my initial reply b/c
> we typically get the SHA1 for free in that case.

As of r1778923, Subversion /trunk always uses the latest authz config
for any HTTP request.

-- Stefan^2.
Received on 2017-01-15 15:06:36 CET

This is an archived mail posted to the Subversion Dev mailing list.