July 2012 Archives by subject

• Messages sorted by: [ thread ] [ author ] [ date ]
• More info on this list...

Starting: Mon Jul 2 10:22:30 CEST 2012
Ending: Tue Jul 31 12:42:46 CEST 2012
Messages: 52

• [3.0] 596246e Make n_sess be the difference between in use and released session objects.   Martin Blix Grydeland
• [3.0] b164c06 Fix for an issue where the bodybytes counter is not updated correctly on gunzipped delivery   Tollef Fog Heen
• [3.0] 1223d71 Changelog for 3.0.3   Andreas Plesner Jacobsen
• [3.0] 1f4006e Merge branch '3.0' of ssh://git.varnish-cache.org/git/varnish-cache into 3.0   Dag Haavi Finstad
• [3.0] 9d09170 Regex set for doing device detection   Lasse Karstensen
• [3.0] a71c7de Make the ban lurker skip busy objects   Martin Blix Grydeland
• [3.0] b50bbd6 Fix for an issue where the bodybytes counter is not updated correctly on gunzipped delivery   Dag Haavi Finstad
• [3.0] d1ba932 Do not consider busy objects as candidates in EXP_NukeOne.   Martin Blix Grydeland
• [3.0] ef93091 Do not try to expire objects that are still busy.   Martin Blix Grydeland
• [master] 0008c89 Sanitize and document the Sess{Open|Close} VSL messages.   Poul-Henning Kamp
• [master] 0c82d95 Distribute sess_close.h too   Tollef Fog Heen
• [master] 24d426d Add code to dump VSL descriptions in RST format   Poul-Henning Kamp
• [master] 27d63c2 Eliminate the "init_done" and listen socket fields from struct sess by moving the VSL'ing of socket endpoints earlier in the processing.   Poul-Henning Kamp
• [master] 47dd5d2 Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Martin Blix Grydeland
• [master] 4c4d6f6 Make sure that all sessions, requests and busyobj's have a unique transaction id 'vxid'   Poul-Henning Kamp
• [master] 673c50d Fix parsing of -T and -S arguments from shared memory in varnishadm when using the -n option.   Martin Blix Grydeland
• [master] 72d3574 Log SessionOpen with VSL() so we can be sure it is always the first VSL record for a session.   Poul-Henning Kamp
• [master] 74e2d3f Seems I have either hit a GCC bug or a lack of imagination in the C-standardisation. (Even money if you ask me...)   Poul-Henning Kamp
• [master] 7aa0548 Turn some descriptions into intended RST format.   Poul-Henning Kamp
• [master] 88c7737 Also update waiter_epoll to new SES_* api.   Poul-Henning Kamp
• [master] 9f1fd24 Hmm, that hid too much, try this instead...   Poul-Henning Kamp
• [master] aa9e09c Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Martin Blix Grydeland
• [master] ad65eb5 Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Martin Blix Grydeland
• [master] c31f908 Add my official "statement of interest" in HTTP/2.0   Poul-Henning Kamp
• [master] d105f98 Document the version requirements of the varnishreplay input file.   Martin Blix Grydeland
• [master] dec9455 Fix for an issue where varnishlog reports an incorrect "Length" entry on gunzipped delivery   Dag Haavi Finstad
• [master] e027388 Remove XXX comment overtaken by implementation.   Poul-Henning Kamp
• [master] e855267 Make sure the user understands that this is about the default configuration   Andreas Plesner Jacobsen
• [master] ef1b8cb Don't hand out zero VXIDs we need that as a magic value.   Poul-Henning Kamp
• [master] f41fbd1 Rename '-x dumprst' to '-x dumprstparam', we may have more RST coming in the future.   Poul-Henning Kamp
• [master] 0c9975c Make the ban lurker skip busy objects   Martin Blix Grydeland
• [master] 116393f Do not try to expire objects that are still busy.   Martin Blix Grydeland
• [master] 1172fc9 Various minor cleanups for code clarity.   Poul-Henning Kamp
• [master] 15330c9 Make the sandbox interface more modular and tell the implementation which subprocess we are locking down.   Poul-Henning Kamp
• [master] 3cc4a72 Introduce an enum to make the states of HTTP reception more readable.   Poul-Henning Kamp
• [master] 5536bb3 Further untangle req/session state engines   Poul-Henning Kamp
• [master] 5d976a5 INSTALL: minor update   Tollef Fog Heen
• [master] 610d3cb I have no idea how I ended up incrementting the fail counter by the size of the request, probably sleep copy&paste ?   Poul-Henning Kamp
• [master] 73d989b Patch from Geoff:   Poul-Henning Kamp
• [master] 83ab485 Front is slightly easier to understand   Lasse Karstensen
• [master] 8d1a7b8 Test commit to check trac hooks   Tollef Fog Heen
• [master] a6419c8 Fix HTC_Reinit() return type   Poul-Henning Kamp
• [master] acc8909 Regex set for doing device detection   Lasse Karstensen
• [master] ae3e5a3 Push allocation of request into the assigned worker thread when we come back from the waiter, to lessen amount of work waiter does.   Poul-Henning Kamp
• [master] c22afc8 Do not consider busy objects as candidates in EXP_NukeOne.   Martin Blix Grydeland
• [master] c288891 Fix the parallell make check   Martin Blix Grydeland
• [master] c34f10e Move the temporary accounting structure from wrk to req and eliminate two SES_Charge() calls that have nothing to charge.   Poul-Henning Kamp
• [master] d5c5d5b Sigh...   Poul-Henning Kamp
• [master] f81aedd More untangling of req vs. session state.   Poul-Henning Kamp
• [master] f9c42ca Extend cli_limit to 48k to make room for the full output of 'param.show -l'.   Martin Blix Grydeland
• [master] feac371 Don't do automatic backend retry if we have received anything from the backend at all. The retry is only (pseudo-)valid when we have not received anything at all, indicating that the backend closed before it got our request.   Poul-Henning Kamp
• [master] ffa30f0 Change our chosen "bad_ip" to 192.0.2.255 which is a broadcast address from RFC5737's TEST-NET-1 and put a comment there to explain.   Poul-Henning Kamp

Last message date: Tue Jul 31 12:42:46 CEST 2012
Archived on: Tue Jul 31 12:42:48 CEST 2012

• Messages sorted by: [ thread ] [ author ] [ date ]
• More info on this list...