hc-httpclient-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Leandro Nunes <a-lnu...@hotels.com>
Subject Re: HttpCacheStorage#getEntry called twice on cache hit
Date Wed, 17 May 2017 10:59:03 GMT
Thanks Oleg and Jon,

Oleg,
I didn’t see a problem with the implementation of the RFC. The result I get back is the
one I was expecting (regarding actually calling the origin for the entity or returning the
cached instance - without making the request to the origin). The problem has to do with calling
the datastore twice (which can pose as a problem specially when using an off-heap datastore).

Jon,
I created a gist where you can see the behaviour I’m talking about (https://gist.github.com/leandronunes85/c8f68ca556ac4121bdd956b2ae29c19f).
In this example I’m using a facebook static resource with “Cache-Control: public,max-age=31536000,immutable”.
In my exact use case I get back this header: "Cache-Control: max-age=600, public”. In both
cases the behaviour is the same. The output from running the application in the gist is:
op=getEntry, key='https://www.facebook.com:443/rsrc.php/v3/y4/r/gf6iHxsw8zm.png'
op=getEntry, key='https://www.facebook.com:443/rsrc.php/v3/y4/r/gf6iHxsw8zm.png'
op=getEntry, key='https://www.facebook.com:443/rsrc.php/v3/y4/r/gf6iHxsw8zm.png'
op=getEntry, key='https://www.facebook.com:443/rsrc.php/v3/y4/r/gf6iHxsw8zm.png'
op=putEntry, key='https://www.facebook.com:443/rsrc.php/v3/y4/r/gf6iHxsw8zm.png'
CACHE_MISS - HTTP/1.1 200 OK - Cache-Control: public,max-age=31536000,immutable
op=getEntry, key='https://www.facebook.com:443/rsrc.php/v3/y4/r/gf6iHxsw8zm.png'
op=getEntry, key='https://www.facebook.com:443/rsrc.php/v3/y4/r/gf6iHxsw8zm.png'
CACHE_HIT - HTTP/1.1 200 OK - Cache-Control: public,max-age=31536000,immutable
op=getEntry, key='https://www.facebook.com:443/rsrc.php/v3/y4/r/gf6iHxsw8zm.png'
op=getEntry, key='https://www.facebook.com:443/rsrc.php/v3/y4/r/gf6iHxsw8zm.png'
CACHE_HIT - HTTP/1.1 200 OK - Cache-Control: public,max-age=31536000,immutable

As you can see the datastore is being asked for the entity twice on cache hit (and 4 times
(!) for cache misses).

Thanks for any help you can provide,
Leandro

On May 17, 2017, at 12:39 AM, Jon Moore <jonm@apache.org<mailto:jonm@apache.org>>
wrote:

Yes, unfortunately I haven't had time to work on the caching module for
several years, so I don't remember all the ins and outs of the
implementation. However, Leandro, perhaps you could post the headers for
the original request, the (cached) response, and then the subsequent
request that gets a cache hit? If so, I can possibly help explain the
behavior.

Jon

On Tue, May 16, 2017 at 2:23 PM, Oleg Kalnichevski <olegk@apache.org<mailto:olegk@apache.org>>
wrote:

On Tue, 2017-05-16 at 11:42 +0000, Leandro Nunes wrote:
I just noticed the caching implementation is calling getEntry method
twice when the entry is cached. I did some debugging and it looks
like the first one is related with flushing invalid entries and the
second call is related with actually serving the request. My question
is wether this is the expected behaviour or is something that has
been overlooked and can/should be addressed?

Thanks,
Leandro

Leandro


HttpClient cache module desperately needs some attention. The best
course of action should be to consult the RFC, fix the behavior if
wrong, raise a PR at Github.

Oleg


---------------------------------------------------------------------
To unsubscribe, e-mail: httpclient-users-unsubscribe@hc.apache.org<mailto:httpclient-users-unsubscribe@hc.apache.org>
For additional commands, e-mail: httpclient-users-help@hc.apache.org<mailto:httpclient-users-help@hc.apache.org>



Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message