August 2011 Archives by thread

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

Starting: Mon Aug 1 11:18:08 UTC 2011
Ending: Wed Aug 31 15:27:55 UTC 2011
Messages: 390

• [experimental-ims] fe7e315 Print 0, not - for zero-byte replies   Geoff Simmons
• [experimental-ims] 3a0d628 Don't assert on sleep failure due to signals in VSL library (some utils expect to receive SIGHUP).   Geoff Simmons
• [experimental-ims] c11ac80 Avoid segfaulting if hitmiss or handling hasn't been set yet   Geoff Simmons
• [experimental-ims] 17df9a6 Fix compilation errors (set-but-unused variables)   Geoff Simmons
• [experimental-ims] ba20513 Fix up syntax for matching in docs   Geoff Simmons
• [experimental-ims] 044141f Remove obsolete bits from varnishlog man page   Geoff Simmons
• [experimental-ims] e6a9571 Use vsb for varnishncsa to prevent partial lines on error in format strings   Geoff Simmons
• [experimental-ims] cbd2874 Handle errors in varnishncsa formats slightly better   Geoff Simmons
• [experimental-ims] 3e86cd8 Make it possible to log arbitrary headers in varnishncsa   Geoff Simmons
• [experimental-ims] 041b457 Add support for logging response headers in varnishncsa   Geoff Simmons
• [experimental-ims] 8d88729 Minor clarification about %{X}i   Geoff Simmons
• [experimental-ims] 61b0dcf Revert "Print 0, not - for zero-byte replies"   Geoff Simmons
• [experimental-ims] 39249ec Fall back to using current time if urandom is not available   Geoff Simmons
• [experimental-ims] 2061b6e Adjust paths in spec file for trunk builds   Geoff Simmons
• [experimental-ims] d7c7c39 Increase http_resp_hdr_len and http_req_hdr_len   Geoff Simmons
• [experimental-ims] 0c5337e Change the response code on overly long request lines from 400 to 413 as per discussions in httpbis WG   Geoff Simmons
• [experimental-ims] 4172f98 Fix a bug in the entity-replacement code in the ESI-parser.   Geoff Simmons
• [experimental-ims] 9f3fbdf Merged conditional backend request feature   Geoff Simmons
• [experimental-ims] ec2c6ca Merged conditional backend request feature   Geoff Simmons
• [experimental-ims] cd2eb9c Merge conditional backend requests with current trunk   Geoff Simmons
• [experimental-ims] d41a659 Merged conditional backend request feature   Geoff Simmons
• [experimental-ims] f1951c4 Seed random() on startup from /dev/urandom so vtmpfile actually returns random file names   Geoff Simmons
• [experimental-ims] 0017c88 Merge conditional backend requests with current trunk   Geoff Simmons
• [experimental-ims] a36f0ed Merge branch 'experimental-ims' of ssh://git.varnish-cache.org/git/varnish-cache into experimental-ims   Geoff Simmons
• [master] 2357a95 Fix a classic programmer bug, spotted by the increasingly eagle-eyed LLVM C-compiler.   Poul-Henning Kamp
• [master] 2f10ef7 Another insufficiently wide memset spotted by LLVM.   Poul-Henning Kamp
• [master] f96ecd7 And a third case of sizeof missing dereference spotted by LLVM   Poul-Henning Kamp
• [master] 8888447 Make VSL_Open not open the shmlog if run with -r   Tollef Fog Heen
• [master] 66101b9 Fix up reading of saved log files   Tollef Fog Heen
• [master] 8d662f9 Only strip out -p to status/killproc for old fedora/RHEL   Tollef Fog Heen
• [master] 2c74419 Minor polish to the malloc allocator:   Poul-Henning Kamp
• [master] beb0c5b Cap the TTL (to param "shortlived") when we use the Transient storage to avoid dropping an object on out of storage conditions.   Poul-Henning Kamp
• [master] b5c9ba0 Be more aggressive in getting rid of threads and slower in adding threads, hopefully fixing c00002   Tollef Fog Heen
• [master] f19fab6 Fix typo in help text   Tollef Fog Heen
• [master] a39f3ee Reset bereq http struct on restart from vcl_miss and vcl_pass   Tollef Fog Heen
• [master] 59a0967 Slam grace and keep to zero if we fall back to a transient object.   Poul-Henning Kamp
• [master] e93ef71 Free the workers busyobj if any on cleanup   Martin Blix Grydeland
• [master] 627964a Make HSH_Deref return the busyobj to the worker for reuse when dereferencing and destroying a busy objcore.   Martin Blix Grydeland
• [master] 53a842b Add test-case for #971   Kristian Lyngstøl
• [master] 97ff8c4 Report dlerror if dlopen fails   Tollef Fog Heen
• [master] 573e86f Don't core dump on negative or execessively large fds in "order mode"   Poul-Henning Kamp
• [master] 8510961 Fix the boundary detection code in vsl.c, slightly more paranoid than DocWilcos proposed fix in #956   Poul-Henning Kamp
• [3.0] c0ec0ef Reference varnish-cli + -n semantics   Tollef Fog Heen
• [3.0] 7be3623 Fix compilation errors (set-but-unused variables)   Tollef Fog Heen
• [3.0] 48bfc6d reference varnish-cli(7)   Tollef Fog Heen
• [3.0] fb1dfce Request specific variables are not available in vcl_init{} and vcl_fini{}   Tollef Fog Heen
• [3.0] b96c0af Use file locking on SHMFILE to indicate this file is currently in use   Tollef Fog Heen
• [3.0] 78055b5 Make all three gzip parameter varnish parameters: gzip_level, gzip_window and gzip_memlevel.   Tollef Fog Heen
• [3.0] 16cbc36 Beef the gzip parameters up to the usual defaults: window=15 and memlevel 8.   Tollef Fog Heen
• [3.0] 5e36c5e Explain what a purge is.   Tollef Fog Heen
• [3.0] 9935d3f do_stream   Tollef Fog Heen
• [3.0] 91921da Don't assert on sleep failure due to signals in VSL library (some utils expect to receive SIGHUP).   Tollef Fog Heen
• [3.0] 0b1d542 Avoid segfaulting if hitmiss or handling hasn't been set yet   Tollef Fog Heen
• [3.0] a4c57ec Fix up syntax for matching in docs   Tollef Fog Heen
• [3.0] 7dabcb5 Remove obsolete bits from varnishlog man page   Tollef Fog Heen
• [3.0] f5248e5 Use vsb for varnishncsa to prevent partial lines on error in format strings   Tollef Fog Heen
• [3.0] a716784 Handle errors in varnishncsa formats slightly better   Tollef Fog Heen
• [3.0] 49c62a5 Make it possible to log arbitrary headers in varnishncsa   Tollef Fog Heen
• [3.0] b312abd Add support for logging response headers in varnishncsa   Tollef Fog Heen
• [3.0] 6645f2a Minor clarification about %{X}i   Tollef Fog Heen
• [3.0] 01d04b5 Fall back to using current time if urandom is not available   Tollef Fog Heen
• [3.0] 24eea3f Increase http_resp_hdr_len and http_req_hdr_len   Tollef Fog Heen
• [master] d9b4655 Overhaul the stevedore statistics, the SMA and SMF interpreted certain fields in different ways.   Poul-Henning Kamp
• [master] f67de46 Markup/typos   Tollef Fog Heen
• [master] 048ad79 Update param docs   Tollef Fog Heen
• [master] b40bb36 Don't zap the size until we have updated statistics   Poul-Henning Kamp
• [master] c1a6faa Move 'age' and 'entered' into struct exp.   Poul-Henning Kamp
• [master] 9765e50 Fiddle the "RFC" SLT_TTL record into submission.   Poul-Henning Kamp
• [master] 9f00f52 Fix typo   Tollef Fog Heen
• [master] d6a5687 Document do_esi   Tollef Fog Heen
• [master] bcc71b9 Add a couple of asserts to make sure we only do ims/range on 200.   Poul-Henning Kamp
• [master] 409308d Move the function which checks if we can do condiitional (304) delivery into the RFC policy module where it belongs.   Poul-Henning Kamp
• [master] d9f5779 Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Poul-Henning Kamp
• [master] 5c8bb04 Don't panic if we can both do conditional (IMS) and stream, prefer IMS, it is less work and less data to transmit.   Poul-Henning Kamp
• [master] 029e95a Test case for #972   Poul-Henning Kamp
• [master] 624ec16 Move the 304/ims processing into the object delivery code, where we also have the 206/range handling.   Poul-Henning Kamp
• [master] 12a3f19 Polish up the unit-test case for the binheap, looking for #967   Poul-Henning Kamp
• [master] a6ddafc Clamp rather than overflow on child indexes when we get to the end of the UINT_MAX items we can support.   Poul-Henning Kamp
• [master] 726d93f Duh! Git committing from the wrong directory doesn't do what you expect:   Poul-Henning Kamp
• [master] e8a63f4 Try to read the silo signature to find the correct address to map the silo into VM. If this fails or we get garbage, the silo will be cleared.   Poul-Henning Kamp
• [master] bc2a0bd ifdef the OFOF development tool   Poul-Henning Kamp
• [master] cd55687 Constify   Poul-Henning Kamp
• [master] 94c870b We always allocate an object entirely inside one stevedore now.   Poul-Henning Kamp
• [master] b231cf2 Fix the test for #971   Kristian Lyngstøl
• [master] 369b5b6 Prompted by HTTPbis and FlexeLint in unison, go over the nhttp and response code stuff and fully incorporate the fact that they are both 16 unsigned now.   Poul-Henning Kamp
• [master] e13e923 Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Poul-Henning Kamp
• [master] 0e3c3ed Merge branch 'master' of git+ssh://git.varnish-cache.org/git/varnish-cache   Kristian Lyngstøl
• [master] 1695f2a Move r00971.vtc to tests.disabled/ and add README   Kristian Lyngstøl
• [master] 3dffeb9 Add (disabled) test-case for #977   Kristian Lyngstøl
• [master] 65f4cf6 Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Per Buer
• [master] 0b23a0f document the transient stevedore   Per Buer
• [master] 47e277b Add emphasis to explain when bans are checked against   Per Buer
• [master] 9bef407 more on bans, the lurker   Per Buer
• [master] 6b39846 the basics on persistent storage. Explained the semantics quickly so people understand why not all objects survive a crash   Per Buer
• [master] 6a59dac more on transient. didn't get it rigth the first time. Thanks to scoof for pointing it out.   Per Buer
• [master] 0caee0a My recent 'discover correct -spersistent mmap address' fix broke 32 bit systems.   Poul-Henning Kamp
• [master] 149c4d4 Implement a consistent retry policy in the random/client/hash director:   Poul-Henning Kamp
• [master] b38682b Make sure the entire waiting list is rushed when an object goes non-busy. Not sure what I thought when I changed it last, but it was clearly not smart thinking.   Poul-Henning Kamp
• [master] 466c078 Oops, didn't get this bit of the #963 fix in.   Poul-Henning Kamp
• [master] 298c7c3 We can not use a lenght based response when we transform (gzip/gunzip) we stream and backend didn't send c-l.   Poul-Henning Kamp
• [master] eab8652 restart in vcl_deliver{} would crash in vcl_fetch{} due to missing cleanup.   Poul-Henning Kamp
• [master] 81010e4 Cleare do_stream on all esi objects, including included objects.   Poul-Henning Kamp
• [master] 4c2945d hit_for_pass in vcl_fetch   Per Buer
• [master] 8766504 std.log missing ()   Per Buer
• [master] 2bedcf2 remove cruft about -o   Per Buer
• [master] 38415a2 Add a "sendhex" verb   Poul-Henning Kamp
• [master] a8481d0 Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Poul-Henning Kamp
• [master] f4084bb pulled one of the last changes on -m (removed -O by accident)   Per Buer
• [master] e416c09 added () to EXAMPLES for std.log   Per Buer
• [master] d4681a6 Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Per Buer
• [3.0] ef07e1d Pass priv2 member instead of objcore pointer, thats all we need.   Tollef Fog Heen
• [3.0] d5ee593 Optimize the objhead layout to avoid packing on 64bit systems.   Tollef Fog Heen
• [3.0] c20bb7d Reduce the http.protover field from double to uint8_t for space reasons.   Tollef Fog Heen
• [3.0] ece423d reduce some fieldwidths and pack struct http to avoid packing.   Tollef Fog Heen
• [3.0] 7392101 Pack struct ws better to avoid padding.   Tollef Fog Heen
• [3.0] f9c460c ifdef offsetoff output in debug.sizeof   Tollef Fog Heen
• [3.0] 22c0814 Move the length field to the front of entries in the vary string.   Tollef Fog Heen
• [3.0] b6e6fe9 Use VCT instead of <ctype.h>. Trust HTTP_GetHdr() to remove leading white space.   Tollef Fog Heen
• [3.0] 26c0049 Overhaul the Vary matching code to make it faster, by assuming (but do not trusting) that all objects under an objhdr have the same Vary string.   Tollef Fog Heen
• [3.0] 8c6ce17 Don't overwrite reference results unless told to   Tollef Fog Heen
• [3.0] 1a43c37 Move the setup/teardown of requests vary prediction space from cache_hash to cache_center.   Tollef Fog Heen
• [3.0] a5e2100 Make sure we neuter the next entry after the one we just created to prevent old junk on the workspace from being taken for valid.   Tollef Fog Heen
• [3.0] 8f451fa Move request vary string from wrk to sess   Tollef Fog Heen
• [3.0] 0459355 Remove unused/unreferenced stuff.   Tollef Fog Heen
• [3.0] 03fe1a4 Don't force a rerun to save reference if we just ran it.   Tollef Fog Heen
• [3.0] e3deca9 Enforce that oc_getobj() does not work on BUSY objcores.   Tollef Fog Heen
• [3.0] 35dc396 Introduce the "busyobj" structure which is only valid for busy objects.   Tollef Fog Heen
• [3.0] 58d36cb Fix a compiler fault if two IP comparisons were on the same line of source code.   Tollef Fog Heen
• [3.0] a3607f0 Reintroduce TTL VCL logging that was lost in commit a21746d23d4047ce209a0c283e12ff684f478b72.   Tollef Fog Heen
• [3.0] c426c9b Split registration and selection of backend poll functions into two different functions.   Tollef Fog Heen
• [3.0] c28afe5 Change the response code on overly long request lines from 400 to 413 as per discussions in httpbis WG   Tollef Fog Heen
• [3.0] 3cd1a4e Fix a bug in the entity-replacement code in the ESI-parser.   Tollef Fog Heen
• [3.0] 879ca02 Another insufficiently wide memset spotted by LLVM.   Tollef Fog Heen
• [3.0] 4dd65b9 Fix a classic programmer bug, spotted by the increasingly eagle-eyed LLVM C-compiler.   Tollef Fog Heen
• [3.0] f535d86 And a third case of sizeof missing dereference spotted by LLVM   Tollef Fog Heen
• [3.0] c3aac58 Make VSL_Open not open the shmlog if run with -r   Tollef Fog Heen
• [3.0] 5d93777 Fix up reading of saved log files   Tollef Fog Heen
• [3.0] 89d2b2b Only strip out -p to status/killproc for old fedora/RHEL   Tollef Fog Heen
• [3.0] 01e4580 Minor polish to the malloc allocator:   Tollef Fog Heen
• [3.0] 24e4a2f Cap the TTL (to param "shortlived") when we use the Transient storage to avoid dropping an object on out of storage conditions.   Tollef Fog Heen
• [3.0] 85a7216 Be more aggressive in getting rid of threads and slower in adding threads, hopefully fixing c00002   Tollef Fog Heen
• [3.0] 105ee6c Fix typo in help text   Tollef Fog Heen
• [3.0] a65cb2e Reset bereq http struct on restart from vcl_miss and vcl_pass   Tollef Fog Heen
• [3.0] 088ad84 Slam grace and keep to zero if we fall back to a transient object.   Tollef Fog Heen
• [3.0] d77db1d Free the workers busyobj if any on cleanup   Tollef Fog Heen
• [3.0] b793875 Make HSH_Deref return the busyobj to the worker for reuse when dereferencing and destroying a busy objcore.   Tollef Fog Heen
• [3.0] ca5859d Report dlerror if dlopen fails   Tollef Fog Heen
• [3.0] 38dd274 Don't core dump on negative or execessively large fds in "order mode"   Tollef Fog Heen
• [3.0] 1a45246 Fix the boundary detection code in vsl.c, slightly more paranoid than DocWilcos proposed fix in #956   Tollef Fog Heen
• [3.0] 213ccb2 Add upgrade notes from 2.1 to 3.0   Tollef Fog Heen
• [3.0] 9cc413e Overhaul the stevedore statistics, the SMA and SMF interpreted certain fields in different ways.   Tollef Fog Heen
• [3.0] 132ad32 Markup/typos   Tollef Fog Heen
• [3.0] dfc056a Update param docs   Tollef Fog Heen
• [3.0] f7a2d82 Don't zap the size until we have updated statistics   Tollef Fog Heen
• [3.0] a3af11c Make sure the entire waiting list is rushed when an object goes non-busy. Not sure what I thought when I changed it last, but it was clearly not smart thinking.   Tollef Fog Heen
• [3.0] 400e1bc Oops, didn't get this bit of the #963 fix in.   Tollef Fog Heen
• [3.0] f900240 Move 'age' and 'entered' into struct exp.   Tollef Fog Heen
• [3.0] b26529d Fiddle the "RFC" SLT_TTL record into submission.   Tollef Fog Heen
• [3.0] cbd6a3f Fix typo   Tollef Fog Heen
• [3.0] f7fd7ef Document do_esi   Tollef Fog Heen
• [3.0] af52295 Add a couple of asserts to make sure we only do ims/range on 200.   Tollef Fog Heen
• [3.0] 93686f8 Move the function which checks if we can do condiitional (304) delivery into the RFC policy module where it belongs.   Tollef Fog Heen
• [3.0] 7a8492b Don't panic if we can both do conditional (IMS) and stream, prefer IMS, it is less work and less data to transmit.   Tollef Fog Heen
• [3.0] f9612d2 Test case for #972   Tollef Fog Heen
• [3.0] b4f70f0 Move the 304/ims processing into the object delivery code, where we also have the 206/range handling.   Tollef Fog Heen
• [3.0] a958b02 Polish up the unit-test case for the binheap, looking for #967   Tollef Fog Heen
• [3.0] add1d82 Clamp rather than overflow on child indexes when we get to the end of the UINT_MAX items we can support.   Tollef Fog Heen
• [3.0] ae07042 ifdef the OFOF development tool   Tollef Fog Heen
• [3.0] 6cba43b Try to read the silo signature to find the correct address to map the silo into VM. If this fails or we get garbage, the silo will be cleared.   Tollef Fog Heen
• [3.0] 1142de7 Constify   Tollef Fog Heen
• [3.0] 6b9fa15 We always allocate an object entirely inside one stevedore now.   Tollef Fog Heen
• [3.0] 9703116 Prompted by HTTPbis and FlexeLint in unison, go over the nhttp and response code stuff and fully incorporate the fact that they are both 16 unsigned now.   Tollef Fog Heen
• [3.0] 4784fe9 Add (disabled) test-case for #977   Tollef Fog Heen
• [3.0] e7ce7d8 Add emphasis to explain when bans are checked against   Tollef Fog Heen
• [3.0] 2edd911 more on bans, the lurker   Tollef Fog Heen
• [3.0] b4b63b4 document the transient stevedore   Tollef Fog Heen
• [3.0] a03088c the basics on persistent storage. Explained the semantics quickly so people understand why not all objects survive a crash   Tollef Fog Heen
• [3.0] 2502e03 Implement a consistent retry policy in the random/client/hash director:   Tollef Fog Heen
• [3.0] 693a89b My recent 'discover correct -spersistent mmap address' fix broke 32 bit systems.   Tollef Fog Heen
• [3.0] 0006ddd We can not use a lenght based response when we transform (gzip/gunzip) we stream and backend didn't send c-l.   Tollef Fog Heen
• [3.0] 6f7c8a7 restart in vcl_deliver{} would crash in vcl_fetch{} due to missing cleanup.   Tollef Fog Heen
• [3.0] 0f8805f Cleare do_stream on all esi objects, including included objects.   Tollef Fog Heen
• [3.0] be6cd99 hit_for_pass in vcl_fetch   Tollef Fog Heen
• [3.0] 6aaefe2 std.log missing ()   Tollef Fog Heen
• [3.0] b6580d8 Add a "sendhex" verb   Tollef Fog Heen
• [master] b54440f If the backend used chunked encoding and sent junk after the gzip data, the thread would go into a spin.   Poul-Henning Kamp
• [master] 6ba82cb Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Poul-Henning Kamp
• [3.0] 62cc4b3 added () to EXAMPLES for std.log   Tollef Fog Heen
• [3.0] e6e34d2 If the backend used chunked encoding and sent junk after the gzip data, the thread would go into a spin.   Tollef Fog Heen
• [master] da5a1b9 Catch up with Tollefs Vmod_Varnish_ABI change.   Poul-Henning Kamp
• [master] 5c5bd71 Add a new paramter "nuke_limit" which controls how many objects we are willing to evict per storage allocation attempt.   Poul-Henning Kamp
• [master] 380e7f2 Properly clean up if we bail out of cnt_error because we cannot allocate an object.   Poul-Henning Kamp
• [master] 7ed5f2b The law of unconsidered consequences strikes again:   Poul-Henning Kamp
• [master] fe331b6 Allow larger bodies in varnishtest   Poul-Henning Kamp
• [master] be1eb03 Try to test the object alloc / LRU / Nuke / beresp.storage hinting   Poul-Henning Kamp
• [master] 44f5700 Lock panic buffer before mangling it   Tollef Fog Heen
• [master] eed1a96 Document 3.0 changes   Tollef Fog Heen
• [master] c6e1e88 An assert to make sure...   Poul-Henning Kamp
• [master] fb3d538 SMA->trim() did not return the trimmed space to the pool causing the SMA to run out of space eventually.   Poul-Henning Kamp
• [master] 9e7c2d4 Test stevedore/LRU/Nuke when the body allocation fails   Poul-Henning Kamp
• [master] 6a2fd27 Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Poul-Henning Kamp
• [master] a552ed2 Update sphinx Makefile to contain the right set of files   Tollef Fog Heen
• [master] d0dce47 Adjust c0004[45] limits to work on 32 bit   Tollef Fog Heen
• [3.0] 615ab48 Check the ABI of VMODs.   Tollef Fog Heen
• [3.0] b509f4d Update sphinx Makefile to contain the right set of files   Tollef Fog Heen
• [3.0] d94b756 Catch up with Tollefs Vmod_Varnish_ABI change.   Tollef Fog Heen
• [3.0] 460b1d1 Add a new paramter "nuke_limit" which controls how many objects we are willing to evict per storage allocation attempt.   Tollef Fog Heen
• [3.0] 4c4ef96 Properly clean up if we bail out of cnt_error because we cannot allocate an object.   Tollef Fog Heen
• [3.0] 5626096 The law of unconsidered consequences strikes again:   Tollef Fog Heen
• [3.0] d244202 Allow larger bodies in varnishtest   Tollef Fog Heen
• [3.0] 8e3bcec Try to test the object alloc / LRU / Nuke / beresp.storage hinting   Tollef Fog Heen
• [3.0] 93234bd Adjust c0004[45] limits to work on 32 bit   Tollef Fog Heen
• [3.0] 0bd09c9 An assert to make sure...   Tollef Fog Heen
• [3.0] b93d462 Lock panic buffer before mangling it   Tollef Fog Heen
• [3.0] 4421577 Document 3.0 changes   Tollef Fog Heen
• [3.0] 8d0c80b SMA->trim() did not return the trimmed space to the pool causing the SMA to run out of space eventually.   Tollef Fog Heen
• [3.0] 4cd6bed Test stevedore/LRU/Nuke when the body allocation fails   Tollef Fog Heen
• [master] 27e1139 Move some asserts up in the code, to catch Kristians panic.   Poul-Henning Kamp
• [master] 555735f Even harder asserts for kristians test   Poul-Henning Kamp
• [master] cd6b2b8 Add various flags to the panic output   Poul-Henning Kamp
• [master] 8c534e2 Be much more bombastic about the per-request flags.   Poul-Henning Kamp
• [3.0] 1360bcd Update configure etc for rc1   Tollef Fog Heen
• [master] a4bc345 Add a "fallback" director as a variant of round-robin.   Poul-Henning Kamp
• [master] c67397f Add note about thread_pool_max   Tollef Fog Heen
• [master] e30bec4 Documentation fixes for 3.0   Tollef Fog Heen
• [master] b229950 Recent changes to the RFC2616 ttl calculation confused what ttl value was actually being used.   Poul-Henning Kamp
• [master] 94139b3 Reduce http_req_size on 32 bit machines   Poul-Henning Kamp
• [master] 37586e0 Use SMA instead in the hope that we can escape what I think is a page-size issue on ppc64   Poul-Henning Kamp
• [master] af353a6 Administrative fixes: - Cleanup whitespace bogons - Make all machine-generated no-copyright notices consistent - Update copyrights   Poul-Henning Kamp
• [3.0] f542ba1 Move some asserts up in the code, to catch Kristians panic.   Tollef Fog Heen
• [3.0] 05b386b Even harder asserts for kristians test   Tollef Fog Heen
• [3.0] 4d87432 Add various flags to the panic output   Tollef Fog Heen
• [3.0] 0d1be0b Be much more bombastic about the per-request flags.   Tollef Fog Heen
• [3.0] 6ace083 Add missing header for string concatention operator   Tollef Fog Heen
• [3.0] 19096c1 Add a "fallback" director as a variant of round-robin.   Tollef Fog Heen
• [3.0] bf81373 Add note about thread_pool_max   Tollef Fog Heen
• [3.0] a09a6bf Documentation fixes for 3.0   Tollef Fog Heen
• [3.0] de5f2e3 Recent changes to the RFC2616 ttl calculation confused what ttl value was actually being used.   Tollef Fog Heen
• [3.0] 7739484 Reduce http_req_size on 32 bit machines   Tollef Fog Heen
• [3.0] 54abd5a Use SMA instead in the hope that we can escape what I think is a page-size issue on ppc64   Tollef Fog Heen
• [master] 0820070 Add rudimentary -X support to varnishncsa   Tollef Fog Heen
• [master] a600a23 Fix time_firstbyte bug in varnishncsa   Tollef Fog Heen
• [master] 35aede3 Don't allow "error" in vcl_init{} and vcl_fini{}, they have no request to error.   Poul-Henning Kamp
• [master] 60c3ecf Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Poul-Henning Kamp
• [master] f2d9ff1 Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Poul-Henning Kamp
• [master] 5b65d5a Document vcl_init{} and vcl_fini{}   Poul-Henning Kamp
• [master] 204198f Don't look at a static version of the VSM during startup.   Poul-Henning Kamp
• [master] 40575c5 Reintroduce ExpKill VSL records.   Poul-Henning Kamp
• [master] 8488a8b Clean up VSM setup if mmap fails   Tollef Fog Heen
• [3.0] 4b47459 Administrative fixes: - Cleanup whitespace bogons - Make all machine-generated no-copyright notices consistent - Update copyrights   Tollef Fog Heen
• [3.0] a47e91d Add rudimentary -X support to varnishncsa   Tollef Fog Heen
• [3.0] 35dc8f5 Fix time_firstbyte bug in varnishncsa   Tollef Fog Heen
• [3.0] 1e4f08a Don't allow "error" in vcl_init{} and vcl_fini{}, they have no request to error.   Tollef Fog Heen
• [3.0] 9ac381b Document vcl_init{} and vcl_fini{}   Tollef Fog Heen
• [3.0] bc6aded Don't look at a static version of the VSM during startup.   Tollef Fog Heen
• [3.0] 5306100 Reintroduce ExpKill VSL records.   Tollef Fog Heen
• [3.0] ceda65d Clean up VSM setup if mmap fails   Tollef Fog Heen
• [3.0] 1207602 Make -x dumprst always available   Tollef Fog Heen
• [3.0] c2346d2 Add missing ABI line check that got lost   Tollef Fog Heen
• [3.0] 805e674 Move varnishd parameter documentation into their own file   Tollef Fog Heen
• [3.0] c9dddd2 Add 3.0.1 changes   Tollef Fog Heen
• [3.0] 6152bf7 Bump version numbers for 3.0.1   Tollef Fog Heen
• [3.0] 277dee3 Document various director retry-mechanisms   Kristian Lyngstøl
• [master] 6f7944c Document various director retry-mechanisms   Kristian Lyngstøl
• [experimental-ims] 2357a95 Fix a classic programmer bug, spotted by the increasingly eagle-eyed LLVM C-compiler.   Geoff Simmons
• [experimental-ims] 2f10ef7 Another insufficiently wide memset spotted by LLVM.   Geoff Simmons
• [experimental-ims] f96ecd7 And a third case of sizeof missing dereference spotted by LLVM   Geoff Simmons
• [experimental-ims] 8888447 Make VSL_Open not open the shmlog if run with -r   Geoff Simmons
• [experimental-ims] 66101b9 Fix up reading of saved log files   Geoff Simmons
• [experimental-ims] 8d662f9 Only strip out -p to status/killproc for old fedora/RHEL   Geoff Simmons
• [experimental-ims] 2c74419 Minor polish to the malloc allocator:   Geoff Simmons
• [experimental-ims] beb0c5b Cap the TTL (to param "shortlived") when we use the Transient storage to avoid dropping an object on out of storage conditions.   Geoff Simmons
• [experimental-ims] b5c9ba0 Be more aggressive in getting rid of threads and slower in adding threads, hopefully fixing c00002   Geoff Simmons
• [experimental-ims] f19fab6 Fix typo in help text   Geoff Simmons
• [experimental-ims] a39f3ee Reset bereq http struct on restart from vcl_miss and vcl_pass   Geoff Simmons
• [experimental-ims] 59a0967 Slam grace and keep to zero if we fall back to a transient object.   Geoff Simmons
• [experimental-ims] e93ef71 Free the workers busyobj if any on cleanup   Geoff Simmons
• [experimental-ims] 627964a Make HSH_Deref return the busyobj to the worker for reuse when dereferencing and destroying a busy objcore.   Geoff Simmons
• [experimental-ims] 53a842b Add test-case for #971   Geoff Simmons
• [experimental-ims] 97ff8c4 Report dlerror if dlopen fails   Geoff Simmons
• [experimental-ims] 8510961 Fix the boundary detection code in vsl.c, slightly more paranoid than DocWilcos proposed fix in #956   Geoff Simmons
• [experimental-ims] 573e86f Don't core dump on negative or execessively large fds in "order mode"   Geoff Simmons
• [experimental-ims] ae94311 Add upgrade notes from 2.1 to 3.0   Geoff Simmons
• [experimental-ims] d9b4655 Overhaul the stevedore statistics, the SMA and SMF interpreted certain fields in different ways.   Geoff Simmons
• [experimental-ims] 048ad79 Update param docs   Geoff Simmons
• [experimental-ims] f67de46 Markup/typos   Geoff Simmons
• [experimental-ims] b40bb36 Don't zap the size until we have updated statistics   Geoff Simmons
• [experimental-ims] c1a6faa Move 'age' and 'entered' into struct exp.   Geoff Simmons
• [experimental-ims] 9765e50 Fiddle the "RFC" SLT_TTL record into submission.   Geoff Simmons
• [experimental-ims] 9f00f52 Fix typo   Geoff Simmons
• [experimental-ims] d6a5687 Document do_esi   Geoff Simmons
• [experimental-ims] bcc71b9 Add a couple of asserts to make sure we only do ims/range on 200.   Geoff Simmons
• [experimental-ims] d9f5779 Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Geoff Simmons
• [experimental-ims] 409308d Move the function which checks if we can do condiitional (304) delivery into the RFC policy module where it belongs.   Geoff Simmons
• [experimental-ims] 5c8bb04 Don't panic if we can both do conditional (IMS) and stream, prefer IMS, it is less work and less data to transmit.   Geoff Simmons
• [experimental-ims] 029e95a Test case for #972   Geoff Simmons
• [experimental-ims] 624ec16 Move the 304/ims processing into the object delivery code, where we also have the 206/range handling.   Geoff Simmons
• [experimental-ims] 12a3f19 Polish up the unit-test case for the binheap, looking for #967   Geoff Simmons
• [experimental-ims] a6ddafc Clamp rather than overflow on child indexes when we get to the end of the UINT_MAX items we can support.   Geoff Simmons
• [experimental-ims] 726d93f Duh! Git committing from the wrong directory doesn't do what you expect:   Geoff Simmons
• [experimental-ims] e8a63f4 Try to read the silo signature to find the correct address to map the silo into VM. If this fails or we get garbage, the silo will be cleared.   Geoff Simmons
• [experimental-ims] bc2a0bd ifdef the OFOF development tool   Geoff Simmons
• [experimental-ims] cd55687 Constify   Geoff Simmons
• [experimental-ims] 94c870b We always allocate an object entirely inside one stevedore now.   Geoff Simmons
• [experimental-ims] b231cf2 Fix the test for #971   Geoff Simmons
• [experimental-ims] 369b5b6 Prompted by HTTPbis and FlexeLint in unison, go over the nhttp and response code stuff and fully incorporate the fact that they are both 16 unsigned now.   Geoff Simmons
• [experimental-ims] e13e923 Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Geoff Simmons
• [experimental-ims] 1695f2a Move r00971.vtc to tests.disabled/ and add README   Geoff Simmons
• [experimental-ims] 0e3c3ed Merge branch 'master' of git+ssh://git.varnish-cache.org/git/varnish-cache   Geoff Simmons
• [experimental-ims] 3dffeb9 Add (disabled) test-case for #977   Geoff Simmons
• [experimental-ims] 0b23a0f document the transient stevedore   Geoff Simmons
• [experimental-ims] 65f4cf6 Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Geoff Simmons
• [experimental-ims] 47e277b Add emphasis to explain when bans are checked against   Geoff Simmons
• [experimental-ims] 9bef407 more on bans, the lurker   Geoff Simmons
• [experimental-ims] 6b39846 the basics on persistent storage. Explained the semantics quickly so people understand why not all objects survive a crash   Geoff Simmons
• [experimental-ims] 6a59dac more on transient. didn't get it rigth the first time. Thanks to scoof for pointing it out.   Geoff Simmons
• [experimental-ims] 0caee0a My recent 'discover correct -spersistent mmap address' fix broke 32 bit systems.   Geoff Simmons
• [experimental-ims] 149c4d4 Implement a consistent retry policy in the random/client/hash director:   Geoff Simmons
• [experimental-ims] b38682b Make sure the entire waiting list is rushed when an object goes non-busy. Not sure what I thought when I changed it last, but it was clearly not smart thinking.   Geoff Simmons
• [experimental-ims] 466c078 Oops, didn't get this bit of the #963 fix in.   Geoff Simmons
• [experimental-ims] 298c7c3 We can not use a lenght based response when we transform (gzip/gunzip) we stream and backend didn't send c-l.   Geoff Simmons
• [experimental-ims] eab8652 restart in vcl_deliver{} would crash in vcl_fetch{} due to missing cleanup.   Geoff Simmons
• [experimental-ims] 81010e4 Cleare do_stream on all esi objects, including included objects.   Geoff Simmons
• [experimental-ims] 4c2945d hit_for_pass in vcl_fetch   Geoff Simmons
• [experimental-ims] 2bedcf2 remove cruft about -o   Geoff Simmons
• [experimental-ims] 8766504 std.log missing ()   Geoff Simmons
• [experimental-ims] 38415a2 Add a "sendhex" verb   Geoff Simmons
• [experimental-ims] a8481d0 Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Geoff Simmons
• [experimental-ims] f4084bb pulled one of the last changes on -m (removed -O by accident)   Geoff Simmons
• [experimental-ims] e416c09 added () to EXAMPLES for std.log   Geoff Simmons
• [experimental-ims] d4681a6 Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Geoff Simmons
• [experimental-ims] b54440f If the backend used chunked encoding and sent junk after the gzip data, the thread would go into a spin.   Geoff Simmons
• [experimental-ims] 6ba82cb Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Geoff Simmons
• [experimental-ims] dca3cfb Check the ABI of VMODs.   Geoff Simmons
• [experimental-ims] da5a1b9 Catch up with Tollefs Vmod_Varnish_ABI change.   Geoff Simmons
• [experimental-ims] 5c5bd71 Add a new paramter "nuke_limit" which controls how many objects we are willing to evict per storage allocation attempt.   Geoff Simmons
• [experimental-ims] 380e7f2 Properly clean up if we bail out of cnt_error because we cannot allocate an object.   Geoff Simmons
• [experimental-ims] 7ed5f2b The law of unconsidered consequences strikes again:   Geoff Simmons
• [experimental-ims] fe331b6 Allow larger bodies in varnishtest   Geoff Simmons
• [experimental-ims] be1eb03 Try to test the object alloc / LRU / Nuke / beresp.storage hinting   Geoff Simmons
• [experimental-ims] c6e1e88 An assert to make sure...   Geoff Simmons
• [experimental-ims] 44f5700 Lock panic buffer before mangling it   Geoff Simmons
• [experimental-ims] eed1a96 Document 3.0 changes   Geoff Simmons
• [experimental-ims] fb3d538 SMA->trim() did not return the trimmed space to the pool causing the SMA to run out of space eventually.   Geoff Simmons
• [experimental-ims] 9e7c2d4 Test stevedore/LRU/Nuke when the body allocation fails   Geoff Simmons
• [experimental-ims] 6a2fd27 Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Geoff Simmons
• [experimental-ims] a552ed2 Update sphinx Makefile to contain the right set of files   Geoff Simmons
• [experimental-ims] d0dce47 Adjust c0004[45] limits to work on 32 bit   Geoff Simmons
• [experimental-ims] 27e1139 Move some asserts up in the code, to catch Kristians panic.   Geoff Simmons
• [experimental-ims] 555735f Even harder asserts for kristians test   Geoff Simmons
• [experimental-ims] cd6b2b8 Add various flags to the panic output   Geoff Simmons
• [experimental-ims] 8c534e2 Be much more bombastic about the per-request flags.   Geoff Simmons
• [experimental-ims] 7dd921d Add missing header for string concatention operator   Geoff Simmons
• [experimental-ims] a4bc345 Add a "fallback" director as a variant of round-robin.   Geoff Simmons
• [experimental-ims] c67397f Add note about thread_pool_max   Geoff Simmons
• [experimental-ims] e30bec4 Documentation fixes for 3.0   Geoff Simmons
• [experimental-ims] b229950 Recent changes to the RFC2616 ttl calculation confused what ttl value was actually being used.   Geoff Simmons
• [experimental-ims] 94139b3 Reduce http_req_size on 32 bit machines   Geoff Simmons
• [experimental-ims] 37586e0 Use SMA instead in the hope that we can escape what I think is a page-size issue on ppc64   Geoff Simmons
• [experimental-ims] af353a6 Administrative fixes: - Cleanup whitespace bogons - Make all machine-generated no-copyright notices consistent - Update copyrights   Geoff Simmons
• [experimental-ims] 0820070 Add rudimentary -X support to varnishncsa   Geoff Simmons
• [experimental-ims] 35aede3 Don't allow "error" in vcl_init{} and vcl_fini{}, they have no request to error.   Geoff Simmons
• [experimental-ims] a600a23 Fix time_firstbyte bug in varnishncsa   Geoff Simmons
• [experimental-ims] f2d9ff1 Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Geoff Simmons
• [experimental-ims] 40575c5 Reintroduce ExpKill VSL records.   Geoff Simmons
• [experimental-ims] 5b65d5a Document vcl_init{} and vcl_fini{}   Geoff Simmons
• [experimental-ims] 60c3ecf Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Geoff Simmons
• [experimental-ims] 204198f Don't look at a static version of the VSM during startup.   Geoff Simmons
• [experimental-ims] 8488a8b Clean up VSM setup if mmap fails   Geoff Simmons
• [experimental-ims] 00efb89 Make -x dumprst always available   Geoff Simmons
• [experimental-ims] 2c581a0 Add missing ABI line check that got lost   Geoff Simmons
• [experimental-ims] 666e4d3 Move varnishd parameter documentation into their own file   Geoff Simmons
• [experimental-ims] 5fb157a Merged conditional backend request feature   Geoff Simmons
• [experimental-ims] b3d7c21 Merged conditional backend request feature   Geoff Simmons
• [experimental-ims] b33301f Add 3.0.1 changes   Geoff Simmons
• [experimental-ims] b23d319 Merge conditional backend requests with current trunk   Geoff Simmons
• [experimental-ims] 9fe1576 Merged conditional backend request feature   Geoff Simmons
• [experimental-ims] ad3b0f5 Merge conditional backend requests with current trunk   Geoff Simmons
• [experimental-ims] 5f7a3e1 Merged conditional backend request feature   Geoff Simmons
• [experimental-ims] 5c19ff9 Merged conditional backend request feature   Geoff Simmons
• [experimental-ims] e8a5f69 Merged conditional backend request feature   Geoff Simmons
• [experimental-ims] 8aa5a7e Merge conditional backend requests with current trunk   Geoff Simmons
• [experimental-ims] bccc1f7 Merged conditional backend request feature   Geoff Simmons
• [experimental-ims] d58cad0 Seed random() on startup from /dev/urandom so vtmpfile actually returns random file names   Geoff Simmons
• [experimental-ims] 5a98340 Merge conditional backend requests with current trunk   Geoff Simmons
• [experimental-ims] 13a0986 Merge backend conditional requests with 3.0.1   Geoff Simmons
• [experimental-ims] 71ee192 Merge backend conditional requests with 3.0.1   Geoff Simmons
• [master] 863c22b Allow relational comparisons on REAL type.   Poul-Henning Kamp
• [master] 16d58af Merge branch 'master' of ssh://git.varnish-cache.org/git/varnish-cache   Poul-Henning Kamp
• [master] c0e2acc When headers are surplus to limits, only log the first 20 char.   Poul-Henning Kamp

Last message date: Wed Aug 31 15:27:55 UTC 2011
Archived on: Wed Feb 11 18:06:19 UTC 2026

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