Florian Forster [Sun, 6 Nov 2016 10:10:54 +0000 (11:10 +0100)]
src/write_riemann_threshold.h: Various fixes.
* Add license header. Simply copied src/write_riemann_threadhold.c's license.
* Make header self-contained. "plugin.h" is needed for data_source_t and
value_list_t.
* Add names to the parameters and write documentation for the only exported
function.
Florian Forster [Sun, 6 Nov 2016 09:34:22 +0000 (10:34 +0100)]
write_riemann plugin: Fix order of includes.
The previous version failed on Solaris with:
In file included from ./daemon/collectd.h:31:0,
from write_riemann.c:36:
./config.h:1619:0: error: "_FILE_OFFSET_BITS" redefined [-Werror]
#define _FILE_OFFSET_BITS 64
^
Marc Fournier [Tue, 1 Nov 2016 10:54:26 +0000 (11:54 +0100)]
configure.ac: make finding java deterministic
The sort order of the `find` command proves to be dependent on external
factors, which makes `./configure` pick different java versions on
different systems, making the whole build non-reproducible (see
https://reproducible-builds.org/ for more details).
Adding a call to `sort`, with the locale forced, fixes this issue.
Sebastian Harl [Sat, 24 Sep 2016 10:55:40 +0000 (12:55 +0200)]
RRDCacheD plugin: Try to reconnect upon failed operations.
The RRD client library does not provide means to check the status of a
connection and there's no way to detect that the connection is broken after,
for example, the daemon restarted. To work around that, try to reconnect if
any client operation fails.
Assume, for example, the config `Key "*/foo"`. This config expects JSON
in the form:
{
"bar": {
"foo": 1337
}
}
If the available JSON is instead:
{
"error_code": 0,
"bar": {
"foo": 1337
}
}
the code will take a look at the zero associated with "error_code" and
determine that a map (with key "foo") is expected instead. Previously
the code would continue, eventually calling `cj_get_type()` which
expects that `key->type` is a valid pointer, resulting in a segmentation
fault.
This patch does three things to ensure that this segmentation fault does
not happen again:
1. `cj_get_type()` checks its argument to make sure it is valid before
dereferencing any pointers.
2. In case a non-map is found when a map is expected, the code will
return instead of limping on.
3. After calling `cj_cb_inc_array_index()`, which may update the key,
make sure that it actually did and that key is valid now.
network plugin: Initialize libgcrypt only when needed.
Previously, libgcrypt was initialized unconditionally in network_init(),
which may cause trouble on some systems. With this patch, gcrypt is
initialized if:
* A client socket as signing or encryption enabled, or
* a server socket has an auth_file configured.
Ruben Kerkhof [Thu, 11 Aug 2016 15:44:42 +0000 (17:44 +0200)]
md: fix deprecation warning with latest glibc
md.c: In function ‘md_process’:
md.c:120:3: warning: ‘__makedev_from_sys_types’ is deprecated:
In the GNU C Library, `makedev' is defined by <sys/sysmacros.h>.
For historical compatibility, it is currently defined by
<sys/types.h> as well, but we plan to remove this soon.
To use `makedev', include <sys/sysmacros.h> directly.
If you did not intend to use a system-defined macro `makedev',
you should #undef it after including <sys/types.h>.
[-Wdeprecated-declarations]
if (st.st_rdev != makedev (MD_MAJOR, minor))
^~
In file included from /usr/include/features.h:397:0,
from /usr/include/bits/libc-header-start.h:33,
from /usr/include/stdio.h:28,
from ./daemon/collectd.h:34,
from md.c:22:
David Gibson [Wed, 10 Aug 2016 10:07:14 +0000 (20:07 +1000)]
Fix conceptual buglet in default_callback
The 'DERIVE' path in default_callback() increments the 'counter' field of
value instead of the 'derive' field. Since those fields have the same
type and offset within the value union this will still work fine, but
AFAICT it's conceptually incorrect. This corrects it.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au> Signed-off-by: Florian Forster <octo@collectd.org>
Florian Forster [Sat, 6 Aug 2016 16:22:56 +0000 (18:22 +0200)]
ping plugin: Let start_thread() return gracefully when thread is already running.
Signaling an error caused the init function to signal an error, causing
the read callback to be unregistered on systems where init is called
more than once.
Sebastian Harl [Thu, 4 Aug 2016 20:15:34 +0000 (22:15 +0200)]
python plugin: Fix conversion of value-lists from Python to C.
This affects data-sets with more than one data-source of different types.
Previously, the type of the first data-source would have been used to convert
all values.
network plugin: Fix heap overflow in parse_packet().
Emilien Gaspar has identified a heap overflow in parse_packet(), the
function used by the network plugin to parse incoming network packets.
This is a vulnerability in collectd, though the scope is not clear at
this point. At the very least specially crafted network packets can be
used to crash the daemon. We can't rule out a potential remote code
execution though.
Pavel Rochnyack [Fri, 13 May 2016 13:20:22 +0000 (19:20 +0600)]
perl plugin: Fixes for #1706
* Fix coredump due to destroying interpreter on threads running perl.
* Fix deadlock when perl_log() is called from perl_init()
Pavel Rochnyack [Fri, 13 May 2016 08:50:31 +0000 (14:50 +0600)]
perl plugin: lock base thread interpreter in perl_init() too.
Avoid race conditions with c_ithread_create() called from threads of already-initialized plugins.
Fix for https://github.com/collectd/collectd/issues/1706.
This hopefully fixes "make check" on Solaris, which currently fails
with:
CC common_test.o
CCLD test_common
Undefined first referenced
symbol in file
kc ./.libs/libcommon.a(common.o)
ld: fatal: symbol referencing errors. No output written to test_common
collect2: error: ld returned 1 exit status
Florian Forster [Thu, 11 Jun 2015 13:35:43 +0000 (14:35 +0100)]
Makefile.am: Link tests with required libraries, too.
"make check" fails on Solaris because libraries are missing:
CCLD test_common
Undefined first referenced
symbol in file
kc ./.libs/libcommon.a(common.o)
kstat_data_lookup ./.libs/libcommon.a(common.o)
kstat_lookup ./.libs/libcommon.a(common.o)
kstat_read ./.libs/libcommon.a(common.o)
getaddrinfo ./.libs/libcommon.a(common.o)
freeaddrinfo ./.libs/libcommon.a(common.o)
gai_strerror ./.libs/libcommon.a(common.o)
ld: fatal: symbol referencing errors. No output written to test_common
collect2: error: ld returned 1 exit status
CCLD test_utils_vl_lookup
Undefined first referenced
symbol in file
kstat_data_lookup daemon/.libs/libcommon.a(common.o)
kstat_lookup daemon/.libs/libcommon.a(common.o)
kstat_read daemon/.libs/libcommon.a(common.o)
getaddrinfo daemon/.libs/libcommon.a(common.o)
freeaddrinfo daemon/.libs/libcommon.a(common.o)
gai_strerror daemon/.libs/libcommon.a(common.o)
ld: fatal: symbol referencing errors. No output written to test_utils_vl_lookup
collect2: error: ld returned 1 exit status
librrd 1.6.0 is now threadsafe and librrd_th is gone.
Since there doesn't seem to be a way to detect that librrd
is threadsafe, use pkg-config to check for 1.6.0 or newer instead.
The logic is now as follows:
Check for librrd >= 1.6.0 with pkg-config
If not found, look for rrd_update_r in librrd_th.
If not found, look for rrd_update in librrd and assume librrd is not
thread safe.
projects / thirdparty / collectd.git / log
