source: bin/varnishd/cache_center.c @ da30ce

Revision Log Mode:


Legend:

Added
Modified
Copied or renamed
Diff Rev Age Author Log Message
(edit) @da30ce   5 years phk Replay of SVN r5766 Unify header handling for gzip, and make …
(edit) @0b06c85   5 years phk Replay of SVN r5764: Retire the old ESI code.
(edit) @da5303   5 years phk Gzip support for ESI is turning into a new implementation of ESI, so bite …
(edit) @1d984c3   5 years phk Flesh out the gzip and gunzip fetch processors. Now you can get varnish …
(edit) @0eb5cc   5 years phk Un-const the sp argument to vpf methods git-svn-id: …
(edit) @51600e   5 years phk VTLATD (VTLA Of The Day :-) VFP, Varnish Fetch Processors, are code that …
(edit) @89d6a3   5 years phk Make sure we have space to align the Content-Length header we create. …
(edit) @9d8a7a2   5 years phk Rename sp->esis to sp->esi_level and make it available in VCL as …
(edit) @6875d5   5 years phk Rework the fetch code, to use a common function for feeding bytes into …
(edit) @ad6d79   5 years phk Rename a variable to better reflect what it really is. git-svn-id: …
(edit) @4b1248   5 years phk Only set chunked mode if we actually want the body (ie: Not HEAD) …
(edit) @ef9c88   5 years phk Further unify delivery path so that gunzip, esi and direct object don't …
(edit) @253cfa   5 years phk Further unravel og unify the response path. git-svn-id: …
(edit) @6a28ea   5 years phk Use sp->wrk->res_mode more. git-svn-id: …
(edit) @c9fb8e   5 years phk Centralize the "how do we deliver this ?" magic into a sp->wrk variable …
(edit) @ac1b99d   5 years phk Remove XXX comments OBE git-svn-id: …
(edit) @5c3619   5 years phk Beat a path through, where we can successfully ungzip an trivial object, …
(edit) @24041b   5 years phk Pass beresp.storage into stevedore selection as a hint, and pick any …
(edit) @70ae38   5 years phk Implement beresp.storage as a string variable which lives entirely inside …
(edit) @1cc33a   5 years phk Remove an old workaround for VCL calls from LRU processing. git-svn-id: …
(edit) @9381351   5 years martin Implement restart in vcl_deliver(), and add test case for it. Fixes: #411
(edit) @ae96ce   5 years phk Make EXP_Touch() responsible for updating the o->last_lru timestamp. …
(edit) @dfd1a0a   5 years phk Use drand48() for vmod_std.random(). Seed the RAND48 generator when we …
(edit) @3e3008f   5 years phk Eliminate sp->objhead as a general variable, it was always tracking …
(edit) @8053cd7   5 years phk Collapse the object allocation/creation to a 1-phase process which …
(edit) @4db9d2   5 years phk This is really complicated to explain, but quite simple behind all the …
(edit) @223fd73   5 years phk Move the oc default methods and HSH_Object() into stevedore.c where it …
(edit) @218de76   5 years phk Move the ban pointer to objcore, as long planned. Clean up assert for …
(edit) @2e849d   5 years phk Rework/simplify the obj/objcore dereference logic. git-svn-id: …
(edit) @8f25fe   5 years phk We are still not fully adapted to the fact that objcore is the main handle …
(edit) @4324b6   5 years phk Reduce the arguments of HSH_DerefObjCore() to the minimum needed. …
(edit) @653d39   5 years phk One of the silly overgeneralizations in RFC2616, is that headers which …
(edit) @3bd239e   5 years phk Change how we do If-Modified-Since on objects without a Last-Modified …
(edit) @147b7c   5 years phk If requests come in fast enough on a single connection, typically when …
(edit) @e708a8   5 years phk Move the determination of the existence of and how we will fetch a body in …
(edit) @9d7b65   5 years phk Isolate the text-representation of the HTTP response status field in …
(edit) @6c2baa   5 years phk Nitpicking style etc. git-svn-id: …
(edit) @365848   5 years martin Introduce req.hash_always_miss (force cache miss) and req.hash_ignore_busy …
(edit) @1e555c   5 years phk white-space anal-retentive fixup git-svn-id: …
(edit) @b3bc68   5 years tfheen Correct embedded dot about obj.* in vcl_fetch The embedded dot claimed …
(edit) @1e5f67   5 years phk Retry backend fetches one time if we got a recycled conncetion and the …
(edit) @b53d64   5 years phk Rename backend connections to the VTLA "VBC" git-svn-id: …
(edit) @a9cad4   5 years phk Add a new VCL variable "client.identity" which defaults to client.ip …
(edit) @8b3a30   5 years kristian Revert the return(refresh) patches (better approach coming soon-ish) …
(edit) @118689   5 years phk Do not suppress the Content-Length:, Age:, Range: and Proxy-Auth headers …
(edit) @8105f4   5 years kristian Return(refresh) from VCL recv, allowing a controlled refresh of content …
(edit) @aefb425   5 years phk Make Expect 100-continue test case-insensitive Candidate for merging into …
(edit) @4b04e3   5 years phk Fix a problem with accounting of ESI transactions, in particular make the …
(edit) @58e0a0   5 years phk Emit Length for client side right before ReqEnd, to summarize ESI …
(edit) @6a8830   5 years phk Split and rename shmlog.h to vsl.h and vsc.h git-svn-id: …
(edit) @fb84ba   5 years phk Whitespace cleanup git-svn-id: …
(edit) @26e8c7   5 years phk We converted HEAD to GET in vcl_recv{}, that is far too early, since it …
(edit) @62e6b5   5 years phk Add a HTTP_Copy() function and use it for the "rollback" copy of req.* …
(edit) @7eba946   5 years phk Stop messing with the linger bit, there are too many portability issues …
(edit) @3e6375   5 years phk Tell FlexeLint? that we ignore returnvalues on purpose. git-svn-id: …
(edit) @8cc7f9   5 years phk Make the session_linger a one-shot timer, and apply it also to new …
(edit) @c08184   5 years phk Update copyright messages of files significantly changed in 2010 so far. …
(edit) @5b251d   5 years phk White-space cleanup git-svn-id: …
(edit) @e78f18   5 years phk Comments with a warning I came close to forgetting git-svn-id: …
(edit) @b39e11   5 years phk Remove obj_workspace parameter. We autosize it for just what is needed. …
(edit) @8c8c7e   6 years phk Move the backend timeouts out of struct sess, they don't belong there. …
(edit) @10c90c   6 years phk Take another trip over the CLI code, this time the help function. …
(edit) @573c98   6 years phk Make the number of HTTP headers a parameter, and allocate only as many as …
(edit) @9aed2f   6 years phk More flexelint silencing git-svn-id: …
(edit) @ed03c7   6 years phk Put the hash-digest into struct sess, let HSH_Lookup grab it from there. …
(edit) @3129d0   6 years phk Execute vcl_hash{} right after vcl_recv{}, so that vcl_pipe{} and …
(edit) @6736b5   6 years phk Eliminate the obj.hash hack. This was originally a quick hack to allow …
(edit) @f06f73   6 years phk White-space cleanup megacommit. git-svn-id: …
(edit) @06e68d8   6 years phk Move the Vary specification into the object worksspace instead of using …
(edit) @b16d7f   6 years phk If we cannot even make sense of the request, don't bother with attempting …
(edit) @203069   6 years phk If we come into vcl_error{} with a non-3-digit error status, for instance …
(edit) @c57175c   6 years phk Polish the object/storage allocation code a bit, in preparation for some …
(edit) @c4b1e7   6 years phk Rename and move HSH_NewObject() to STV_NewObject(), this is a more …
(edit) @6063e2   6 years phk Account for unresurrected persistent objects in their own stats counter to …
(edit) @fbe00d4   6 years phk Originally, we did not have the bereq.* and beresp.* variable sets, and …
(edit) @e75ecf   6 years phk A clarifying comment. git-svn-id: …
(edit) @7d36ca   6 years phk Try to simplify the variable structure around the decision to populate a …
(edit) @8ef8c6   6 years phk More stats work. Add a parameter, so the worker thread gets forced to …
(edit) @8bd6aa9   6 years phk Work over the stats while I wait for -spersistent to crash: Move most of …
(edit) @299b3b8   6 years phk Do the simple part of ban list lurker: link the objcores off the bans the …
(edit) @4323eb   6 years sky it was possible to have infinite recursion between vcl_error returning …
(edit) @a34567   6 years phk Move the objhead reference from the object to the objcore, we need it for …
(edit) @f58652f   6 years phk "Hit-for-pass" objects are not transient. git-svn-id: …
(edit) @79a782   6 years des Embed RCS tag in each compiled file. git-svn-id: …
(edit) @9c7a5c   6 years sky tiny nit git-svn-id: …
(edit) @6a0e902   6 years phk typo git-svn-id: …
(edit) @6832de   6 years phk Make session_linger the default with 50msec timeout. Change the session …
(edit) @48d13a   6 years phk Take another stab at #487 by explicitly resetting the worker thread …
(edit) @97b7b3   6 years phk We need to reset the worker threads workspace between requests if we …
(edit) @1c62f6   6 years phk Be much more consistent about snapshotting session workspace when doing an …
(edit) @c5980b   6 years phk Quiet flexelint down a bit. git-svn-id: …
(edit) @4c351b   6 years sky Allows you to turn off esi processing for a specific request using set …
(edit) @c45586   6 years phk Check that object has a ban before we unbusy it. git-svn-id: …
(edit) @e37eba   6 years phk r4046 forgot to reset SO_LINGER for pipe handling which basically broke …
(edit) @7243e9   6 years phk Change the way we close client sessions. Previously we always used …
(edit) @9392fc   6 years phk Previously, we forced the request to "GET" on all fetches for cache and …
(edit) @366bb3   6 years phk Only make object uncacheable if both obj.ttl and obj.grace is zero. …
(edit) @abe538c   6 years kristian Allow restart in vcl_error git-svn-id: …
(edit) @874530   6 years phk Remember to reset sp->objcore when object inherits reference. …
(edit) @d7f66f   6 years phk Add a new HSH_Insert() function which does that. Clean up the hash …
Note: See TracRevisionLog for help on using the revision log.