X-Git-Url: https://git.octo.it/?a=blobdiff_plain;f=src%2Fcollectd.conf.pod;h=2f3322fb9f5e193742990aed15d9f6f61c9c6420;hb=cd0d4b77c043f7ef118c817aa900bfc7e6c1f8cc;hp=6f5ee7fb0b2e7ad3b0efa6bcf385c3d7da182045;hpb=35b0e3f1704c9a6811443d7a4326337d94023cd9;p=collectd.git

diff --git a/src/collectd.conf.pod b/src/collectd.conf.pod
index 6f5ee7fb..2f3322fb 100644
--- a/src/collectd.conf.pod
+++ b/src/collectd.conf.pod
@@ -3210,6 +3210,30 @@ Pause to apply between attempts of connection to gpsd in seconds (default 5 sec)
 
 =back
 
+=head2 Plugin C<gpu_nvidia>
+
+Efficiently collects various statistics from the system's NVIDIA GPUs using the
+NVML library. Currently collected are fan speed, core temperature, percent
+load, percent memory used, compute and memory frequencies, and power
+consumption.
+
+=over 4
+
+=item B<GPUIndex>
+
+If one or more of these options is specified, only GPUs at that index (as
+determined by nvidia-utils through I<nvidia-smi>) have statistics collected.
+If no instance of this option is specified, all GPUs are monitored.
+
+=item B<IgnoreSelected>
+
+If set to true, all detected GPUs B<except> the ones at indices specified by
+B<GPUIndex> entries are collected. For greater clarity, setting IgnoreSelected
+without any GPUIndex directives will result in B<no> statistics being
+collected.
+
+=back
+
 =head2 Plugin C<grpc>
 
 The I<grpc> plugin provides an RPC interface to submit values to or query
@@ -8841,6 +8865,33 @@ dynamic number assigned by the kernel. Otherwise, C<coreE<lt>nE<gt>> is used
 if there is only one package and C<pkgE<lt>nE<gt>-coreE<lt>mE<gt>> if there is
 more than one, where I<n> is the n-th core of package I<m>.
 
+=item B<RestoreAffinityPolicy> I<AllCPUs>|I<Restore>
+
+Reading data from CPU has side-effect: collectd process's CPU affinity mask
+changes. After reading data is completed, affinity mask needs to be restored.
+This option allows to set restore policy.
+
+B<AllCPUs> (the default): Restore the affinity by setting affinity to any/all
+CPUs.
+
+B<Restore>: Save affinity using sched_getaffinity() before reading data and
+restore it after.
+
+On some systems, sched_getaffinity() will fail due to inconsistency of the CPU
+set size between userspace and kernel. In these cases plugin will detect the
+unsuccessful call and fail with an error, preventing data collection.
+Most of configurations does not need to save affinity as Collectd process is
+allowed to run on any/all available CPUs.
+
+If you need to save and restore affinity and get errors like 'Unable to save
+the CPU affinity', setting 'possible_cpus' kernel boot option may also help.
+
+See following links for details:
+
+L<https://github.com/collectd/collectd/issues/1593>
+L<https://sourceware.org/bugzilla/show_bug.cgi?id=15630>
+L<https://bugzilla.kernel.org/show_bug.cgi?id=151821>
+
 =back
 
 =head2 Plugin C<unixsock>
@@ -9279,12 +9330,6 @@ You can also specify combinations of the B<name> and B<uuid> fields.
 For example B<name uuid> means to concatenate the guest name and UUID
 (with a literal colon character between, thus I<"foo:1234-1234-1234-1234">).
 
-=item B<Instances> B<integer>
-
-How many read instances you want to use for this plugin. The default is one,
-and the sensible setting is a multiple of the B<ReadThreads> value.
-If you are not sure, just use the default setting.
-
 =item B<ExtraStats> B<string>
 
 Report additional extra statistics. The default is no extra statistics, preserving
@@ -9335,11 +9380,38 @@ B<Note>: I<perf> metrics can't be collected if I<intel_rdt> plugin is enabled.
 =back
 
 =item B<PersistentNotification> B<true>|B<false>
+
 Override default configuration to only send notifications when there is a change
 in the lifecycle state of a domain. When set to true notifications will be sent
 for every read cycle. Default is false. Does not affect the stats being
 dispatched.
 
+=item B<Instances> B<integer>
+
+How many read instances you want to use for this plugin. The default is one,
+and the sensible setting is a multiple of the B<ReadThreads> value.
+
+This option is only useful then domains are specially tagged.
+If you are not sure, just use the default setting.
+
+The reader instance will only query the domains with attached a matching tag.
+Tags should have the form of 'virt-X' where X is reader instance number,
+starting from 0.
+
+The special-purpose reader instance #0, guaranteed to be always present,
+will query all the domains with missing or unrecognized tag, so no domain will
+ever left out.
+
+Domain tagging is done with custom attribute in the libvirt domain metadata
+section. Value is taken by XPath I</domain/metadata/ovirtmap/tag/text()>
+expression in I<http://ovirt.org/ovirtmap/tag/1.0> namespace.
+(XPath and namespace values are not configurable yet).
+
+Tagging could be used by management application to evenly spread the
+load among the reader threads, or to pin on the same threads all
+the libvirt domains which use the same shared storage, to minimize
+the disruption in presence of storage outages.
+
 =back
 
 =head2 Plugin C<vmem>
@@ -9673,6 +9745,13 @@ B<Options:>
 
 =over 4
 
+=item B<Host> I<Host>
+
+Bind to the hostname / address I<Host>. By default, the plugin will bind to the
+"any" address, i.e. accept packets sent to any of the hosts addresses.
+
+This option is supported only for libmicrohttpd newer than 0.9.0.
+
 =item B<Port> I<Port>
 
 Port the embedded webserver should listen on. Defaults to B<9103>.