Removed some legacy calls to DEBUG().
Signed-off-by: Sebastian Harl <sh@tokkee.org>
Signed-off-by: Florian Forster <octo@huhu.verplant.org>
Signed-off-by: Sebastian Harl <sh@tokkee.org>
Signed-off-by: Florian Forster <octo@huhu.verplant.org>
rrdtool plugin: Use `-b' (begin) when creating rrd-files.
When receiving data from a host that has a clock that lacks behind the servers
clock, the newly created rrd-files had a timestamp that was ten seconds in the
past from the server's point of view. If the client was, for example, 10:00
minutes late, there would therefore be 9:50 minutes of invalid data which cause
warnings, empty graphs and overall missery. This commit changes this behavior
to not use the default `begin' time but set it explicitely based on the
client's time.
When receiving data from a host that has a clock that lacks behind the servers
clock, the newly created rrd-files had a timestamp that was ten seconds in the
past from the server's point of view. If the client was, for example, 10:00
minutes late, there would therefore be 9:50 minutes of invalid data which cause
warnings, empty graphs and overall missery. This commit changes this behavior
to not use the default `begin' time but set it explicitely based on the
client's time.
ChangeLog: Documented the fix in the apcups plugin.
Merge branch 'uni/collectd-4.0' into collectd-4.0
apcups plugin: Fix the types used in the plugin.
They still had an `apcups_' prefix which isn't legal anymore.
They still had an `apcups_' prefix which isn't legal anymore.
collectd: Make `mkdir' error messages more identifiable.
ChangeLog: Fix a merge bug.
ChangeLog: Added a note about the improved `sstrerror' function.
contrib/collection.cgi: Fix a bug in the `meta_graph_generic_stack' function.
Also add a meta-graph-handler for if_[rt]x_errors.
Also add a meta-graph-handler for if_[rt]x_errors.
Merge branch 'pull/collectd-4.0' into collectd-4.0
sstrerror: Be even more cautios with the return value of `strerror_r'..
I don't trust this GNU implementation one bit.. ;)
I don't trust this GNU implementation one bit.. ;)
contrib/collection.cgi: Display MySQL traffic as bits/s, not bytes/s.
contrib/collection.cgi: Added meta graph handlers for nfs_procedure, ps_state, swap, mysql_commands, and mysql_handler.
src/common.c: Copy the error-string to the buffer - no matter which implementation is used.
contrib/collection.cgi: Don't show all the plugins on the host's page.
Waiting for so many graphs to load is no fun. You _have_ to specify the plugins
you're interested in now.
Waiting for so many graphs to load is no fun. You _have_ to specify the plugins
you're interested in now.
contrib/collection.cgi: Quite some changes, but still incomplete and not very usable..
Merge branch 'pull/collectd-4.0' into collectd-4.0
Added the AC_FUNC_STRERROR_R check.
Even though Posix requires "strerror_r" to return an "int", some systems
(e.g. the GNU libc) return a "char *" _and_ ignore the second argument.
sstrerror() (in src/common.c) has been changed to be aware of this and
handle both cases correctly.
Signed-off-by: Sebastian Harl <sh@tokkee.org>
Signed-off-by: Florian Forster <octo@huhu.verplant.org>
Even though Posix requires "strerror_r" to return an "int", some systems
(e.g. the GNU libc) return a "char *" _and_ ignore the second argument.
sstrerror() (in src/common.c) has been changed to be aware of this and
handle both cases correctly.
Signed-off-by: Sebastian Harl <sh@tokkee.org>
Signed-off-by: Florian Forster <octo@huhu.verplant.org>
src/plugin.c: Remove a legacy debug message.
collectd.conf(5): Undo the changes of the previous merge.
This problem doesn't exist in this version.
This problem doesn't exist in this version.
Merge branch 'collectd-3.11' into collectd-4.0
Conflicts:
ChangeLog
configure.in
Conflicts:
ChangeLog
configure.in
Bumped version to 3.11.6; Updated ChangeLog.
ChangeLog: Added a short note about the fix in the processes plugin.
mysql plugin: Use a different define to determine if the read functionality should be enabled.
collectd.conf(5): Added a note about that trailing whitespace bug.
Bumped version to 4.0.7; Updated ChangeLog.
Merge branch 'collectd-3.11' into collectd-4.0
processes plugin: Fix a possible segfault.
build system: Several fixes for libcurl, libmysql, and libupsclient.
libcurl: Check for CFLAGS using `curl-config --cflags' and use these
CFLAGS when checking for <curl/curl.h> to allow for weird
locations of the library.
libmysql: Cleaned up the checking code.
libupsclient: Check if the header file provides `UPSCONN' or `UPSCONN_t' and
use the one available. This should fix compile issues with with
the nut plugin.
libcurl: Check for CFLAGS using `curl-config --cflags' and use these
CFLAGS when checking for <curl/curl.h> to allow for weird
locations of the library.
libmysql: Cleaned up the checking code.
libupsclient: Check if the header file provides `UPSCONN' or `UPSCONN_t' and
use the one available. This should fix compile issues with with
the nut plugin.
build system: Use the libs provided by `mysql_config' to check for the lib itself.
Merge branch 'pull/collectd-4.0' into collectd-4.0
build system: Use `mysql_config' to find the appropriate flags for MySQL.
contrib/collectd2html.pl: Don't invoke the Perl interpreter with `-w'.
17 years agocontrib/collectd2html.pl: Allow generation of graphs from remote hosts or in an alter...
contrib/collectd2html.pl: Allow generation of graphs from remote hosts or in an alternative directory.
In the contribs area of the package there is a nice collectd2html.pl script
which assumes that the data is present in /var/lib/collectd and infers the
hostname. All fine and dandy, but there is no way to override this. This is
useful if one generates the data on a machine, but generates the graphs on
another one.
With the attached patch now is possible to specify a different host and/or a
different directory for the data.
The added options are '--host-is' and '--data-dir'.
An invocation for a remote host whose rrd files have been placed in
/var/tmp/collectd-remoteh looks like:
collectd2html.pl --host-is=remoteh --data-dir=/var/tmp/collectd-remoteh
In the contribs area of the package there is a nice collectd2html.pl script
which assumes that the data is present in /var/lib/collectd and infers the
hostname. All fine and dandy, but there is no way to override this. This is
useful if one generates the data on a machine, but generates the graphs on
another one.
With the attached patch now is possible to specify a different host and/or a
different directory for the data.
The added options are '--host-is' and '--data-dir'.
An invocation for a remote host whose rrd files have been placed in
/var/tmp/collectd-remoteh looks like:
collectd2html.pl --host-is=remoteh --data-dir=/var/tmp/collectd-remoteh
Merge branch 'pull/collectd-4.0' into collectd-4.0
Fixed some typos in the manpages.
The adjective "performant" does not seem to exist (neither in English nor
in German). I've replaced it with "efficient".
Signed-off-by: Sebastian Harl <sh@tokkee.org>
Signed-off-by: Florian Forster <octo@huhu.verplant.org>
The adjective "performant" does not seem to exist (neither in English nor
in German). I've replaced it with "efficient".
Signed-off-by: Sebastian Harl <sh@tokkee.org>
Signed-off-by: Florian Forster <octo@huhu.verplant.org>
README, contrib/README: Fixed some typos.
Signed-off-by: Sebastian Harl <sh@tokkee.org>
Signed-off-by: Florian Forster <octo@huhu.verplant.org>
Signed-off-by: Sebastian Harl <sh@tokkee.org>
Signed-off-by: Florian Forster <octo@huhu.verplant.org>
network plugin: Allow port arguments to be numeric.
README: Added a pointer to the mailinglist.
contrib/fedora/collectd.spec: Updated for 4.0.6.
Hi,
Here is an updated version of the F7 rpm spec file.
It now handles updates as well (rpm -U).
When new versions of collectd are out please bump the version tag at the top of
the spec file.
It will then build correct version of collectd.
Regards
Kjell Randa
Signed-off-by: Florian Forster <octo@huhu.verplant.org>
Hi,
Here is an updated version of the F7 rpm spec file.
It now handles updates as well (rpm -U).
When new versions of collectd are out please bump the version tag at the top of
the spec file.
It will then build correct version of collectd.
Regards
Kjell Randa
Signed-off-by: Florian Forster <octo@huhu.verplant.org>
collectd(1): Fix typo.
Bumped version to 4.0.6; Updated ChangeLog.
Merge branch 'pull/collectd-4.0' into collectd-4.0
src/collectd.conf.in: Added the `SocketFile' option for the email plugin.
Fixed Fedora specfile for Fedora 7 and collectd 4.0.5
Hi,
Here is a working spec file for Fedora 7.
Copy the tar.gz file to /usr/src/redhat/SOURCES and run
rpmbuild -ba collectd.spec
It has only been tested on x86_64, but I well see if I can find time to test
on x86 as well.
Upgrade not tested.
The file /etc/collection.conf referenced in collection.cgi is missing in the
tar file and not installed by the spec file.
Regards
Kjell Randa
Hi,
Here is a working spec file for Fedora 7.
Copy the tar.gz file to /usr/src/redhat/SOURCES and run
rpmbuild -ba collectd.spec
It has only been tested on x86_64, but I well see if I can find time to test
on x86 as well.
Upgrade not tested.
The file /etc/collection.conf referenced in collection.cgi is missing in the
tar file and not installed by the spec file.
Regards
Kjell Randa
syslog plugin: Call `openlog' right when the plugin is loaded.
sensors plugin: Fix the use of the ignorelist, so it matches the documentation.
Bumped version to 4.0.5; Updated ChangeLog.
collectd-nagios(1): Fix typo.
dns plugin: Added support for the old bind-interface.
Apparently it is still in use, even in new distributions:
-- 8< --
bsbuild>uname -a
OpenBSD bsbuild 4.1 GENERIC#1435 i386
bsbuild>/usr/sbin/named -v
BIND 9.3.4
-- >8 --
Apparently it is still in use, even in new distributions:
-- 8< --
bsbuild>uname -a
OpenBSD bsbuild 4.1 GENERIC#1435 i386
bsbuild>/usr/sbin/named -v
BIND 9.3.4
-- >8 --
src/common.c: Allow the `pwbufp' argument to `getpwnam_r' to be NULL.
Impelemt an own version of `getpwnam_r'.
Since OpenBSD doesn't provide a threadsafe version itself, we need to provide
it in this case.. :/
Since OpenBSD doesn't provide a threadsafe version itself, we need to provide
it in this case.. :/
{hddtemp,mbmon,ntpd} plugins: Only use the `AI_ADDRCONFIG' if it's defined.
configure.in: Include `sys/types.h' and `sys/param.h' before other `sys/*.h' includes.
This fixes build problems under OpenBSD.
This fixes build problems under OpenBSD.
Bumped version to 4.0.4; Updated ChangeLog.
src/collectd-nagios.c: Copy the NAN-defines from `collectd.h'.
This should fix the bug reported by Daniel Leite.
This should fix the bug reported by Daniel Leite.
dns plugin: Check for `net/ppp_defs.h' and include it before `net/if_ppp.h' if it exists.
This hopefully solves build problems under OpenBSD.
This hopefully solves build problems under OpenBSD.
memory plugin: Fix a typo in the Solaris code.
liboconfig: Added `%option noyywrap' to the scanner.
This should solve the link issues when collectd is built without flex being
installed.
This should solve the link issues when collectd is built without flex being
installed.
load plugin: Fix the Linux code that reads `/proc/loadavg'.
cpu, interface plugin: Fix compilation issues under Solaris.
Bumped version to 4.0.3.
ChangeLog, README: Documented the `--with-nan-emulation' option.
configure: Implemented `--with-nan-emulation' for crosscompiling.
When crosscompiling you cannot do the runtime checks for the (0.0 / 0.0) hack.
With this option you can force this implementation if neccessary.
When crosscompiling you cannot do the runtime checks for the (0.0 / 0.0) hack.
With this option you can force this implementation if neccessary.
src/Makefile.am: When building outside the source tree, install will fail
Signed-off-by: Ezra Peisach <epeisach@med-xtal.bu.edu>
Signed-off-by: Ezra Peisach <epeisach@med-xtal.bu.edu>
csv, rrdtool plugin: Initialize a buffer correctly.
While reading the source code, I found the following:
In csv.c (line 44) and rrdtool.c (line 363) (value_list_to_string) the
following syntax exists:
memset (buffer, '\0', sizeof (buffer_len));
The sizeof should be removed - assuming that the desire is to zero the buffer
and not the first four (sizeof(int)) bytes.
Based on the code paths - I do not see an information leak - or any security
implications...
While reading the source code, I found the following:
In csv.c (line 44) and rrdtool.c (line 363) (value_list_to_string) the
following syntax exists:
memset (buffer, '\0', sizeof (buffer_len));
The sizeof should be removed - assuming that the desire is to zero the buffer
and not the first four (sizeof(int)) bytes.
Based on the code paths - I do not see an information leak - or any security
implications...
ChangeLog: Added changes for 4.0.3.
ping plugin: Actually check the uid, not the address of `getuid'.
cpu plugin: Submit the right values in the Darwin code.
Bumped version to 4.0.2; Updated ChangeLog.
hddtemp, ntpd plugin: Allow numeric `Port' arguments.
Bumped version to 4.0.1; Updated ChangeLog.
contrib/collection.cgi: Added graphs for gauge, if_errors, ipt_bytes and ipt_packets.
README, collectd.conf(5): its/it's in documentation
A grammatical pet peeve of mine...
Cheers,
Evan
A grammatical pet peeve of mine...
Cheers,
Evan
collectd.conf.pod: Fixed a typo.
The rrdtool plugin does not create CSV-files ;-)
Signed-off-by: Sebastian Harl <sh@tokkee.org>
The rrdtool plugin does not create CSV-files ;-)
Signed-off-by: Sebastian Harl <sh@tokkee.org>
iptables plugin: Fix a segfault when a comment is passed in the configuration.
Bumped version to 4.0.0; Updated ChangeLog.
Marked the perl plugin to be experimental.
Signed-off-by: Sebastian Harl <sh@tokkee.org>
Signed-off-by: Sebastian Harl <sh@tokkee.org>
Added Perl module "Collectd" to contrib.
Signed-off-by: Sebastian Harl <sh@tokkee.org>
Signed-off-by: Sebastian Harl <sh@tokkee.org>
Merge branch 'pull/master'
collectd-unixsock(5): Corrected the information about the PUTVAL format.
collectd-exec(5): Documented the recent changes.
exec plugin: Change the input format so it's identical to the one used in unixsock.
unixsock plugin: Moved the `parse_identifier' code to `common.c'
The new function `parse_values' in `common.c' can be used to parse
`time:value0:value1:...', as used by RRDTool, the unixsock plugin and, in the
very near future, the exec plugin.
The new function `parse_values' in `common.c' can be used to parse
`time:value0:value1:...', as used by RRDTool, the unixsock plugin and, in the
very near future, the exec plugin.
ntpd plugin: Replace `syslog (LOG_ERR, ...)' with `ERROR (...)'.
Merge branch 'collectd-3.11'
Conflicts:
ChangeLog
configure.in
Conflicts:
ChangeLog
configure.in
Bumped version to 3.11.5; Updated ChangeLog.
ntpd plugin: Fix a possible buffer overflow.
Merge branch 'pull/master'
network plugin: Prepend the plugin's name to debug messages.
exec plugin: Print a warning when the returned string is incorrect.
Also the debug messages have been beautified.
Also the debug messages have been beautified.
contrib/migrate-3-4.px: Added usage information.
collectd.c: Added '-t' command line option.
This option allows to test the configuration only. The program immediately
exits after parsing the config file. A return code not equal to zero indicates
an error.
Some typos have been fixed as well.
Signed-off-by: Sebastian Harl <sh@tokkee.org>
This option allows to test the configuration only. The program immediately
exits after parsing the config file. A return code not equal to zero indicates
an error.
Some typos have been fixed as well.
Signed-off-by: Sebastian Harl <sh@tokkee.org>
memory plugin: Fix unused variable warning under FreeBSD.
Merge branch 'collectd-3.11'
Conflicts:
src/processes.c
Conflicts:
src/processes.c
processes plugin: ps_read_tasks will fail under Linux 2.4.
Take this into account and don't report all processes without tasks as zombies.
Take this into account and don't report all processes without tasks as zombies.
sensors plugin: Let the read-function fail if something isn't right..
..so it gets suspended as it should be.
..so it gets suspended as it should be.
17 years agoplugin.c: Added a comment that explains why the order of the linked list and callback...
plugin.c: Added a comment that explains why the order of the linked list and callback are important.
plugin.c: Advance the llentry_t pointer before calling the shutdown callback.
> Important note: Unregistering the shutdown callback during shutdown is
> somewhat tricky. Doing that frees the llist_t entry in plugin.c:list_shutdown.
> This might cause a segfault in plugin_shutdown_all () when it tries to access
> the next pointer of that entry. If plugin_unregister_shutdown () is the last
> statement before the return from the shutdown callback this should be save in
> about 99% of all cases. Still there should be some better way to handle this.
This is an imho nice way to fix it.
Signed-off-by: Sebastian Harl <sh@tokkee.org>
> Important note: Unregistering the shutdown callback during shutdown is
> somewhat tricky. Doing that frees the llist_t entry in plugin.c:list_shutdown.
> This might cause a segfault in plugin_shutdown_all () when it tries to access
> the next pointer of that entry. If plugin_unregister_shutdown () is the last
> statement before the return from the shutdown callback this should be save in
> about 99% of all cases. Still there should be some better way to handle this.
This is an imho nice way to fix it.
Signed-off-by: Sebastian Harl <sh@tokkee.org>
perl plugin: Unregister all callbacks from collectd when shutting down.
If we do not unregister the callbacks there is still an entry point into the
plugin after Perl has been shut down and its memory has been freed. This will
cause segmentation faults if collectd tries to call any of the registered
callbacks, e.g. by some plugin calling plugin_log () during shutdown.
Important note: Unregistering the shutdown callback during shutdown is
somewhat tricky. Doing that frees the llist_t entry in plugin.c:list_shutdown.
This might cause a segfault in plugin_shutdown_all () when it tries to access
the next pointer of that entry. If plugin_unregister_shutdown () is the last
statement before the return from the shutdown callback this should be save in
about 99% of all cases. Still there should be some better way to handle this.
Signed-off-by: Sebastian Harl <sh@tokkee.org>
If we do not unregister the callbacks there is still an entry point into the
plugin after Perl has been shut down and its memory has been freed. This will
cause segmentation faults if collectd tries to call any of the registered
callbacks, e.g. by some plugin calling plugin_log () during shutdown.
Important note: Unregistering the shutdown callback during shutdown is
somewhat tricky. Doing that frees the llist_t entry in plugin.c:list_shutdown.
This might cause a segfault in plugin_shutdown_all () when it tries to access
the next pointer of that entry. If plugin_unregister_shutdown () is the last
statement before the return from the shutdown callback this should be save in
about 99% of all cases. Still there should be some better way to handle this.
Signed-off-by: Sebastian Harl <sh@tokkee.org>