1
commit 8bf13c304b6ab5fc1a372b49d55c78370efe14ca
2
Author: Balazs Scheidler <bazsi@balabit.hu>
3
Date: Mon Aug 2 17:26:10 2010 +0200
7
commit dc7d07b5881c4f6c94e6f747f4a44779d42b1d20
9
Author: Balazs Scheidler <bazsi@balabit.hu>
10
Date: Fri Jul 9 11:27:07 2010 +0200
12
Merge remote branch '3.0/master'
19
commit c88009fc97b5014e330ecc8ead747691a05b7e97
20
Author: Balazs Scheidler <bazsi@balabit.hu>
21
Date: Fri Jul 9 10:26:06 2010 +0200
23
preparations for 3.0.8
25
commit 6d1ce4e5bdb517941235f059b556fb6dc9d45363
26
Author: Balazs Scheidler <bazsi@balabit.hu>
27
Date: Fri Jul 9 10:25:31 2010 +0200
29
csv-parser: fixed drop-invalid handling
31
drop-invalid was not handled properly and caused valid messages
34
The parser drops messages if:
35
* not all columns are processed
36
* there are leftover characters at the end of the parsing
38
Greedy columns can be empty though.
40
Testcases were extended to hopefully cover all cases now.
42
Reported-By: andrewn at locus.net
44
commit 836be625d8ab1cfcbf26069bc9cb7b2f083d3fdf
45
Merge: 89fecc1 9c2c2be
46
Author: Balazs Scheidler <bazsi@balabit.hu>
47
Date: Tue Jul 6 18:00:33 2010 +0200
49
Merge remote branch '3.0/master'
53
tests/unit/test_csvparser.c
54
tests/unit/test_template_speed.c
56
commit 89fecc1529fce074b5339be334f7ab0425bf8b82
57
Author: Balazs Scheidler <bazsi@balabit.hu>
58
Date: Sat Jul 3 12:37:13 2010 +0200
60
pdbtool: added setlocale in order to accept accented characters in the command line
62
commit 9c2c2be8b95add3789a91cec25ea92a3bd6ceedb
63
Author: Balazs Scheidler <bazsi@balabit.hu>
64
Date: Wed Jun 30 17:41:06 2010 +0200
66
logwriter: fix memory leak if the suppressed message doesn't fit into the queue
68
In case a suppressed message couldn't be inserted into the destination FIFO,
69
it was leaked. This patch correctly disposes of the message in this case.
71
Reported-By: Andy Kitchingman
73
commit 674a569c470dd01435c044c4b26ce4afe000d3fa
74
Author: Balazs Scheidler <bazsi@balabit.hu>
75
Date: Tue Jun 29 17:41:57 2010 +0200
77
test_template_speed: decrease the number of iterations
79
in order to make the unit test go faster.
81
commit b6ff55b0955cfe285ac7afe2063f832daed078a9
82
Author: Balazs Scheidler <bazsi@balabit.hu>
83
Date: Tue Jun 29 17:41:36 2010 +0200
85
csv-parser: fixed greedy + drop-invalid combination
87
greedy & drop-invalid never really worked well in combination, if they both
88
were specified, the message always got rejected. This patch fixes this
91
commit 8c87164ac40257aceaba37ebbc4987c7868ac251
92
Author: Balazs Scheidler <bazsi@balabit.hu>
93
Date: Mon Jun 21 20:04:59 2010 +0200
95
affile: fixed flush_timeout() processing for destination files
97
The patch "logwriter: fixed flush_timeout handling on non-file destinations"
98
broke flush_timeout() processing for file destination, this patch fixes that.
100
commit 1abff81e8da4cb0d767a0d7f7f5f7e36ceb43ab4
101
Merge: 92bdb39 d85c015
102
Author: Balazs Scheidler <bazsi@balabit.hu>
103
Date: Thu Jun 17 08:57:15 2010 +0200
105
Merge remote branch '3.0-local/master'
111
commit d85c0159fc6cd1dcab2ba19439406e29042b6cf7
112
Author: Balazs Scheidler <bazsi@balabit.hu>
113
Date: Sun Jun 6 13:49:52 2010 +0200
115
preparations for 3.0.7
117
commit ed1363ac6c36f6d91a43523eff7ed7ac6a017f9b
118
Author: Balazs Scheidler <bazsi@balabit.hu>
119
Date: Sun Jun 6 13:24:48 2010 +0200
121
logwriter: fixed flush_timeout handling on non-file destinations
123
Non-file destinations may not always be willing to accept messages, which
124
means that even if the flush_timeout elapsed, the destination may not be
125
ready to accept messages.
127
However log_writer_fd_prepare() basically assumed that once the timeout
128
elapses we can immediately start writing messages (which is true for
129
file destinations but false for others).
131
This patch fixes this case.
133
Kudos to Evan who helped me to track down this issue.
135
Reported-By: Evan Rempel
137
commit 5f82516db500209df040215c39da1b5adfc4d143
138
Author: Balazs Scheidler <bazsi@balabit.hu>
139
Date: Mon May 10 17:50:22 2010 +0200
141
main: stay in foreground if syntax-check or debug is requested
143
There's simply no reason to go into the background just to check the
144
syntax of the configuration file.
146
Likewise, debug messages may generate feedback loops in the configuration
147
thus it is not safe to enable them unless internal messages are only
148
going to the stderr and syslog-ng is configured to be run in
151
Reported-By: Corey Hickey
153
commit 92bdb39b2b2236deaf3e5915261419928dc16c79
154
Author: Balazs Scheidler <bazsi@balabit.hu>
155
Date: Sun May 9 13:32:58 2010 +0200
157
Revert "logmsg: don't expect a hostname for local transports in the log message"
159
This reverts commit d8e06f304206aaa9b9e9e62defc6d4f4cfab1505.
161
commit 13de2c5d1c2d4a6b7af7f7bb72eee7ba103405ca
162
Author: Balazs Scheidler <bazsi@balabit.hu>
163
Date: Sat May 8 14:19:56 2010 +0200
165
loggen: added SSL support and the ability to read messages from a file
167
commit 595538c4b55a5e96b5de9f31fac037025679b548
168
Author: Balazs Scheidler <bazsi@balabit.hu>
169
Date: Sat May 8 14:04:58 2010 +0200
171
[loggen] added -q (--quiet) option to loggen, to avoid clobbering the build output
173
commit 4406ecfbf6b5f1f89b46b2b4d1c6a8f1398e4301
174
Author: Tevesz Andras <ghost@balabit.hu>
175
Date: Mon Mar 8 11:13:40 2010 +0100
177
fixed global configuration object possible use after free problem (fixes: #nobug)
179
commit 11cfa6cdec9073b7c0d33adba0733422d872449a
180
Author: Tevesz Andras <ghost@balabit.hu>
181
Date: Fri Feb 26 15:46:56 2010 +0100
183
added upstart syslog-ng config file
185
commit bd85edb19f9f8f787b10162ab531b125aa5e7038
186
Author: Balazs Scheidler <bazsi@balabit.hu>
187
Date: Sat May 8 13:52:33 2010 +0200
189
[unitests] added patterndb v3 unit test (fixes: #nobug)
191
commit 9db07a83047cc6a665f2152ae8c16cd9e5a669a4
192
Author: Balazs Scheidler <bazsi@balabit.hu>
193
Date: Sat May 8 13:47:59 2010 +0200
195
[tests] renamed patterndb unit test to reflect the patterndb version
197
commit e742d478f4a9ac7cb53f31893ea167c24a4e4e07
198
Author: Tevesz Andras <ghost@balabit.hu>
199
Date: Tue Feb 23 10:05:19 2010 +0100
201
[patterdb] fixed patterdb reload function (fixes: #TK-06589)
203
If the new pattern db file is not loadable,
204
then syslog-ng will not drop the old one.
206
commit d2bf600b0565412b2737755a640cfab1c7140f8c
207
Author: Balazs Scheidler <bazsi@balabit.hu>
208
Date: Sat May 8 13:27:36 2010 +0200
210
[control socket] the control socket can set with a parameter
212
commit 01808d5accf701979c9410d9aed07ea71449c25d
213
Author: Balazs Scheidler <bazsi@balabit.hu>
214
Date: Sat May 8 12:45:12 2010 +0200
216
update-patterndb: new command to generate the patterndb from a set of patterndb snippets
218
The update-patterndb script can be used to generate the pattern database
219
file to be used by syslog-ng based on a set of patterndb snippets stored in
220
/etc/syslog-ng/patterndb.d
222
commit 36ec94f2acd0bf53124c89b92955c57d71bbee43
223
Author: Juhasz Viktor <jviktor@balabit.hu>
224
Date: Wed Feb 10 11:39:23 2010 +0100
226
[test_performance] use tail -n +1 instead of tail -1 because this doesn't work in hpux-11
229
commit f8a2de9a7aea5b6b8d9923e57f453592103d18ab
230
Author: Balazs Scheidler <bazsi@balabit.hu>
231
Date: Fri May 7 09:52:18 2010 +0200
233
remove all syslog.h references
235
commit e7b7bb1aa43a9ee8e477c1f425a951d4f5965382
236
Author: Balazs Scheidler <bazsi@balabit.hu>
237
Date: Sat May 8 13:48:46 2010 +0200
239
test_resolve_pwgr: fixed compilation warnings
241
.../tests/unit/test_resolve_pwgr.c: In function ‘main’:
242
.../syslog-ng-ose/tests/unit/test_resolve_pwgr.c:8: warning: pointer targets in passing argument 2 of ‘resolve_user’ differ in signedness
244
commit 644d3453ccd050a9fb929a90ba0174c967a02b19
245
Merge: d8e06f3 6113023
246
Author: Balazs Scheidler <bazsi@balabit.hu>
247
Date: Wed May 5 12:04:41 2010 +0200
249
Merge commit '3.0-local/master'
255
tests/unit/Makefile.am
257
commit 6113023acd0ebb3a91c98e7d733952afb12e598d
258
Author: Balazs Scheidler <bazsi@balabit.hu>
259
Date: Wed May 5 11:38:48 2010 +0200
261
introduce a new syntax to skip owner/group/perm setting on a per-file basis
263
This change was originally an intended behaviour, but due to a design
264
problem how global options were inherited to a per-destination basis
265
at the end it wasn't.
267
The documented behaviour to skip owner/group/perm setting on a given file
268
is to use "-1" as the value, however this only gets applied if _both_
269
the per-destination and the global values are "-1".
271
E.g. if you want something like:
273
options { owner("adm"); };
275
destination d_file_def { file("/var/log/default-owner"); };
276
destination d_file { file("/var/log/no-owner" owner("-1")); };
278
The intention with a config like this is ambigious:
279
* the logical outcome would be to tell syslog-ng not to touch the
280
ownership of the "no-owner" file in any way
281
* the reality however is that "adm" ownership would be applied to
282
"no-owner" as well, since the "-1" value is used internally in syslog-ng
283
to indicate that no per-destination value was specified.
285
The only way to tell syslog-ng not to touch the ownership of a file is to
286
set the global setting to -1 and either set -1 for the destination, or omit
287
the setting (which defaults to -1).
289
Since it would be tricky to fix this without incompatibilities, I've decided
290
not to make any, and introduced an additional syntax, which I think is
291
easier to understand than the original trick with "-1":
293
destination d_file { file("/var/log/no-owner" owner()); };
295
This means that the file ownership shouldn't be changed regardless of the
296
global setting. Internally this is equal to "-2", so the above is equivalent to
298
destination d_file { file("/var/log/no-owner" owner(-2)); };
300
But that's an implementation detail, you don't need to know.
302
Please note that this patch depends on the parent patch as well, without
303
it, it won't function properly.
305
commit 8eac35f1a66d7775a00e309ad0e7b1a745e2e608
306
Author: Balazs Scheidler <bazsi@balabit.hu>
307
Date: Wed May 5 11:27:40 2010 +0200
309
Make sure that negative values never get resolved using getpwnam/getgrnam
311
Originally reported by the SUSE maintainer, syslog-ng is sometimes resolving
312
invalid usernames using getpwnam/getgrnam. Before this patch,
313
syslog-ng first attempted to resolve usernames using NSS, then if that failed
314
tried to interpret them as numeric values.
316
This logic was reversed, now we first check if it is a number, then if that
317
fails do we proceed with resolving the username using getpwnam.
319
Once here, I've also did some cleanups to make the bed for a follow-up patch:
320
* change all uid_t/gid_t/mode_t variables to "gint" instead of their
321
respective types, to make sure that we can represent negative values
323
* all occurences which tried to interpret uid/gid/mode information was
324
changed to interpret all negative values the same as -1 (do nothing).
325
This makes it possible to use further extremal values in the future.
327
The original report came with a patch, but the original goal
328
was completely reimplemented using a different means by Balazs Scheidler.
330
Reported-By: Marius Tomaschewski
332
commit cb6f0452fc5748fdfa64fdbd5ec01f41cb83170f
333
Author: Balazs Scheidler <bazsi@balabit.hu>
334
Date: Wed May 5 10:10:05 2010 +0200
336
SIGHUP handling: block SIGHUP while reload is being carried out
338
Reported by the SUSE maintainer that it is possible for syslog-ng to receive
339
multiple SIGHUP signals while the reload operation is being carried out.
341
This patch changes SIGHUP/SIGCHLD processing to block these signals while
342
handling the signal is being done.
344
Reported-By: Marius Tomaschewski
346
commit d8e06f304206aaa9b9e9e62defc6d4f4cfab1505
347
Author: Balazs Scheidler <bazsi@balabit.hu>
348
Date: Mon May 3 16:51:52 2010 +0200
350
logmsg: don't expect a hostname for local transports in the log message
352
syslog-ng expected the same message format for both local and network
353
transports. However this seems to cause problems for applications that
354
embed a space in their $PROGRAM field, e.g.
356
logger -t "alma korte" message
358
Since we know that $HOST is always empty for local log transports
359
(checked for Linux, FreeBSD, AIX, HP-UX and Solaris) we can simply
360
assume that even if there's a space it is not a hostname, but rather
361
an incorrectly formatted $PROGRAM.
363
This will be recognized as:
369
E.g. if it is formatted with the default template, it becomes the same as it
372
Reported-By: Marius Tomaschewski
374
commit 29ce812d5047021eb0269996df389d1d6b250cde
375
Author: Balazs Scheidler <bazsi@balabit.hu>
376
Date: Mon May 3 16:43:35 2010 +0200
378
logmsg: always store the legacy message header (MSGHDR)
380
This patch essentially makes the 'store-legacy-msghdr' behaviour
381
the default and creates a new 'dont-store-legacy-msghdr' flag to return
382
to the 3.0 compatible behaviour.
384
This patch introduces some incompatibility, however I feel that since the
385
'store-legacy-msghdr' flag is the assumed behaviour, it is best to make it
386
default. The only reason it existed in 3.0 is the expected performance
387
drop, but the 3.1 NVTable refactorizations solved that issue.
389
It is also needed for the follow-up patch to work properly.
391
commit bf76c1bef8a62d5eec7f59ec22349aba5d931c0b
392
Author: Balazs Scheidler <bazsi@balabit.hu>
393
Date: Mon May 3 10:02:43 2010 +0200
395
don't cache the local hostname accross SIGHUPs
397
Reported by the SUSE maintainer, syslog-ng caches
398
the local hostname accross SIGHUPs, which causes problems when
399
dynamic hostname configuration is in use (e.g. DHCP). This patch
400
relookups the local hostname when a SIGHUP is received.
402
Also, there was a bug in the per-source use_fqdn() processing which
403
is fixed by this patch: earlier, if two sources used use_fqdn()
404
differently the one which received the first message won and the
405
hostname was set according to its settings, even if the message
406
was received on the other source.
1
408
commit 8747d74491eac3fdf5052194f47a68e659633ca9
2
409
Author: Balazs Scheidler <bazsi@balabit.hu>
3
410
Date: Sun Apr 11 15:52:46 2010 +0200