- 09 Sep, 2016 2 commits
-
-
Pål Hermunn Johansen authored
In the usage section of varnishstat, some non-literal options were not caught by the backport.
-
Nils Goroll authored
In usage/documentation, mark those options with <> which are not a literal string, but rather a placeholder for some value. Conflicts: bin/varnishstat/varnishstat_options.h
-
- 08 Sep, 2016 1 commit
-
-
Dridi Boukelmoune authored
-
- 07 Sep, 2016 1 commit
-
-
Pål Hermunn Johansen authored
This is an adaptation of b2c18706 by Federico G. Schwindt. Previously you needed an empty argument (''), which didn't play very well along sysv init scripts and was not documented. Now both "none" and the empty argument work, but an extra warning is echoed if the empty argument is used. The documentation only mentions "none" for disabling the secret, since this is the only way of doing it in the next version. Fixes #2007.
-
- 06 Sep, 2016 1 commit
-
-
Guillaume Quintard authored
fix #2027
-
- 05 Sep, 2016 3 commits
-
-
Dridi Boukelmoune authored
-
Poul-Henning Kamp authored
-
Poul-Henning Kamp authored
-
- 29 Aug, 2016 2 commits
-
-
Dridi Boukelmoune authored
-
Dridi Boukelmoune authored
In addition to harcoded features in varnishtest, this opens a window for out-of-tree uses of the test framework to skip test cases if an external component (a database system, OS-specific capabilities, an environment variable, a library, etc) is missing. This feature takes an extra argument, a command-line that must exit with a zero status. Complex feature testing can nicely be wrapped in scripts at the user's discretion: feature cmd "my --command=line" If the test is skipped, it is logged as: ** top 0.0 === feature cmd "my --command=line" * top 0.0 SKIPPING test, missing feature: my --command=line If the command-line is missing, it is logged as: ** top 0.0 === feature cmd ---- top 0.0 Missing the command line
-
- 13 Aug, 2016 1 commit
-
-
Federico G. Schwindt authored
There are some changes in master that need tlc before backporting. This is one of them.
-
- 12 Aug, 2016 3 commits
-
-
Federico G. Schwindt authored
We will need it if the body was consumed due to a call to std.cache_req_body (VRB_Cache). Analysis and test by daghf. Fixes #1881. Conflicts: bin/varnishd/cache/cache_req_fsm.c
-
Nils Goroll authored
-
Martin Blix Grydeland authored
For some (out of workspace or stevedore allocationi failure) errors VFP_Close would fail to be called when ending up not retrying the fetch. This would then potentially leak VFP entry context data like gzip buffers. Fix this by always calling VFP_Close() when changing fetch state when VFP_Open() has been called. Also VFP_Close() would for some retries be called twice, causing double accounting log records print out. This happens when a late error is caught and streaming is disabled, allowing a retry to be attempted. Fix this by not unconditionally calling VFP_Close() in the error state.
-
- 10 Aug, 2016 2 commits
-
-
Pål Hermunn Johansen authored
-
Nils Goroll authored
We're only holding a read lock on the director, but we're updating the nxt member concurrently. This should be acceptable as a performance tradeoff - the only consequence is that round-robin is not strictly going around - it may occasionally skip a backend or hand out the same multiple times in a row. the race is: thread code A: rr->nxt %= rr->vd->n_backend; // rr->nxt == rr->vd->n_backend - 1 B: rr->nxt++; // rr->nxt == rr->vd->n_backend A: be = rr->vd->backend[nxt]; // BOOM should fix #2024
-
- 09 Aug, 2016 1 commit
-
-
Guillaume Quintard authored
-
- 27 Jul, 2016 1 commit
-
-
Pål Hermunn Johansen authored
The child_poker lives in the manager process, and sends pings to the child. With this patch we check that we actually get a PONG back, and not some random data. If problems are detected, we kill the child. Related to: #2010
-
- 08 Jul, 2016 2 commits
-
-
Dridi Boukelmoune authored
-
Dridi Boukelmoune authored
Fixes #2011
-
- 06 Jul, 2016 2 commits
-
-
Lasse Karstensen authored
-
Lasse Karstensen authored
-
- 30 Jun, 2016 1 commit
-
-
Poul-Henning Kamp authored
Spotted by: Régis Leroy
-
- 28 Jun, 2016 3 commits
-
-
Lasse Karstensen authored
-
Lasse Karstensen authored
Ref: #1994
-
Federico G. Schwindt authored
Fixes #1994.
-
- 27 Jun, 2016 4 commits
-
-
Lasse Karstensen authored
Also fix incorrect-but-working links to issues. Cut&paste. Ref: #1933
-
Poul-Henning Kamp authored
Some time ago I decided that it was more convenient if it were actually possible to locate the compiled VCL shared library based on the loaded VCL name, and removed the part of the subdirectory name which made it unique. Bad idea. Add a comment to make sure I don't get that Idea again. Fixes #1933 Merge notes: * Removed label check, as that doesn't exist in 4.1 * Rewrote commit title. Conflicts: bin/varnishd/mgt/mgt_vcl.c
-
Federico G. Schwindt authored
Conflicts: bin/varnishtest/tests/c00075.vtc doc/sphinx/phk/farfaraway.rst
-
Lasse Karstensen authored
-
- 16 Jun, 2016 5 commits
-
-
Federico G. Schwindt authored
Test by scn@. OK by @phk. Fixes #1963.
-
Pål Hermunn Johansen authored
The counter g_cooling variable is now decremented when memory from the cooling list is freed, making the counter vsm_cooling update accordingly. Fixes: #1984
-
Pål Hermunn Johansen authored
This new parameter specifies how long freed VSM memory is on the cooling list before it is actually freed.
-
Martin Blix Grydeland authored
If an object marked OF_IMSCAND is used as the template during backend IMS revalidation, and the new object also is marked OC_F_PASS (hit-for-pass) in vcl_backend_response, the resulting object would (though ObjCopyAttr) have both OF_IMSCAND and OC_F_PASS. The object could then be used again for backend IMS revalidation (since it has the OF_IMSCAND flag), which would cause troubles as the body would by this time be deleted. Fix by not considering objects marked OC_F_PASS as a stale object candidate during lookup. Fixes: #1956
-
Lasse Karstensen authored
-
- 15 Jun, 2016 5 commits
-
-
Poul-Henning Kamp authored
-
Federico G. Schwindt authored
-
Lasse Karstensen authored
-
Lasse Karstensen authored
-
Martin Blix Grydeland authored
An object failing fetch that is waited upon by the ESI delivery code would cause asserts finding an empty object. If waiting for an object, check the status of the object and deliver zero bytes if failed. Fixes: #1953
-