Hello Amos, I understand what you wrote to me but I really do not have any rule that tells squid to cache wwww.facebook.com header .. I only used refresh_pattern to match Pictures , Videos & certain extensions by using ignore-must-revalidate , ignore-no-store , ignore-no-cache , store-stale .. etc and howcome this rule doesnt work ? refresh_pattern -i \.(htm|html|jhtml|mhtml|php)(\?.*|$) 0 0% This rule tells squid not to cache these extensions if we had static URL or dynamic URL. As I noticed every time you open a website for example www.mtv.com.lb then you try to open it again next day but you get the same news ( yesterday) which confused me and allow me to think that maybe Squid ignore all headers related to website if you cached for example pictures and multimedia objects thats why I was asking which rule might be affecting websites?. I cant spend my time on adding list to "cache deny" on websites that were being cached so I thought of only removing the rule caused squid to cached Websites . How to ignore www.facebook.com not to cache but at the same time I want to cache pictures , FLV Videos , CSS , JS but not the header of the main page (HTML/PHP). refresh_pattern ^http:\/\/www\.facebook\.com$ 0 0% 0 I tried to use $ after .com as I only wanted not to cache the main page of Facebook but still I want to cache Pictures and Videos at Facebook and so on at other websites . Sorry If I didnt explain well . Ghassan On Sat, Nov 12, 2011 at 1:47 PM, Amos Jeffries <squid3@xxxxxxxxxxxxx> wrote: > On 12/11/2011 10:30 a.m., Ghassan Gharabli wrote: >> >> Hello, >> >> I am facing a trouble with Caching HTTP Headers. >> >> Everyday I see that www.facebook.com header is being cached and then I >> try to remove it manually from Cache and so other websites ... > > ...other websites what? > >> >> I tried to add these refresh_patterns before any rule but >> unfortunately with no luck! > > Okay some basics. This is a bit complex so if I'm not clear please mention. > > There are several algorithms affecting caching. > > Firstly is absolute expiry. > This tells Squid exactly when to erase the object. Down to the second. > Controlled by Expires: header, or a Cache-Control header with private, > no-store, max-age= values. > As Squid HTTP/1.1 support increases Expires: (a HTTP/1.0 feature) is > getting ignored more often. > > Secondly, there are freshness algorithm. > This tells Squid exactly when the object can be used immediately, or needs > revalidation before use. It is an estimation only. > Controlled by the Date, Last-Modified headers, with Cache-Control > max-stale, etc mixed in as well. > This is where refresh_pattern happens, its min/pct/max values are used to > set boundaries in the decision about staleness. The wiki and refresh_pattern > config docs cover exactly how it works, so I wont repeat it all here. > > > Thirdly, there are the variant algorithm(s). > These tell Squid whether the object in cache is relevant to the request at > all or needs to be skipped. Controlled by the ETag, Vary, Accept, etc. > > To complicate things refresh_pattern has ignore-* an doverride-* options > which make Squid ignore the particular header. These are mostly HTTP > violations and can prevent immediate expiry or extend the estimation well > beyond anything that woudl otherwise be chosen. > NOTE: all these options and refresh_pattern itself can only *extend* the > time something is cached for. They cannot and do not prevent caching or > remove things early. refresh_pattern can have the appearance of shortening > cache times, *if* and only if, the object was caused to be cached that long > by another refresh_pattern estimation later down the list (ie our default 1 > week storage time in the "." pattern line). > > To prevent caching use "cache deny ...". > > >> >> refresh_pattern -i \.(htm|html|jhtml|mhtml|php)(\?.*|$) 0 0% >> 0 >> refresh_pattern ^http:\/\/www\.facebook\.com$ 0 0% 0 >> >> REFRESH_PATTERN CONFIG : >> ------------------------------------------------ >> # 1 year = 525600 mins, 1 month = 43800 mins >> refresh_pattern >> (get_video|videoplayback|videodownload|\.flv).*(begin|start)\=[1-9][0-9]* >> 0 >> 0% 0 >> refresh_pattern -i \.(htm|html|jhtml|mhtml|php)(\?.*|$) 0 0% >> 0 >> refresh_pattern ^http:\/\/www\.facebook\.com$ 0 0% 0 > > This rule above will never match. Squid uses absolute URLs to pass the > refresh pattern. > Absolute URL have a "/" in the path. *never* will ".com" be the last four > characters of the absolute URL. > > Amos >
