[squid-users] Squid logs TCP_MISS/200 for a served cached object requested with If-None-Match

[Garri Djavadyan]

On Sat, 2016-11-19 at 01:12 +0500, Garri Djavadyan wrote:
> Hello,
> 
> I noticed that Squid logs TCP_MISS/200 when it serves previously
> cached 
> object in return to non-matched conditional request with If-None-
> Match. 
> For example:
> 
> 1. Non-conditional request to the previously cached object.
> 
> $ curl -v -x http://127.0.0.1:3128 
> http://mirror.comnet.uz/centos/7/os/x86_64/GPL >/dev/null
> 
> < HTTP/1.1 200 OK
> < Server: nginx
> < Date: Fri, 18 Nov 2016 19:58:38 GMT
> < Content-Type: application/octet-stream
> < Content-Length: 18009
> < Last-Modified: Wed, 09 Dec 2015 22:35:46 GMT
> < ETag: "5668acc2-4659"
> < Accept-Ranges: bytes
> < Age: 383
> < X-Cache: HIT from gentoo.comnet.uz
> < Via: 1.1 gentoo.comnet.uz (squid/5.0.0-BZR)
> < Connection: keep-alive
> 
> 
> 2. Conditional request with non-matching entity to the same object.
> 
> $ curl -v -x http://127.0.0.1:3128 -H 'If-None-Match: "5668acc2-
> 4658"' 
> http://mirror.comnet.uz/centos/7/os/x86_64/GPL >/dev/null
> 
> < HTTP/1.1 200 OK
> < Server: nginx
> < Date: Fri, 18 Nov 2016 19:58:38 GMT
> < Content-Type: application/octet-stream
> < Content-Length: 18009
> < Last-Modified: Wed, 09 Dec 2015 22:35:46 GMT
> < ETag: "5668acc2-4659"
> < Accept-Ranges: bytes
> < X-Cache: MISS from gentoo.comnet.uz
> < Via: 1.1 gentoo.comnet.uz (squid/5.0.0-BZR)
> < Connection: keep-alive
> 
> 
> I found that the behavior is related to the following code 
> (client_side_reply.cc):
> 
>          if (!e->hasIfNoneMatchEtag(r)) {
>              // RFC 2616: ignore IMS if If-None-Match did not match
>              r.flags.ims = false;
>              r.ims = -1;
>              r.imslen = 0;
>              r.header.delById(Http::HdrType::IF_MODIFIED_SINCE);
> --->        http->logType = LOG_TCP_MISS;
>              sendMoreData(result);
>              return true;
>          }
> 
> 
> So, it seems like intended behavior, but I can't understand the
> reasons.
> Or maybe it is a bug?
> 
> Thanks.
> 
> Garri


Any comments will be much appreciated. Thanks.

Garri