]>
git.ipfire.org Git - thirdparty/collectd.git/log
Matwey V. Kornilov [Fri, 7 Feb 2025 15:46:58 +0000 (18:46 +0300)]
Fix deprecated actions/upload-artifact@v3 in GitHub Actions
(cherry picked from commit
0edf8505c239b021dead7b1226b71db131cae692 )
WinstonMDP [Wed, 23 Oct 2024 17:29:13 +0000 (20:29 +0300)]
Format
WinstonMDP [Wed, 23 Oct 2024 16:59:57 +0000 (19:59 +0300)]
Fix lcc_receive
The unit has been substracted to pass gcc stringop-truncation warning in
lcc_receive function
Matwey V. Kornilov [Thu, 22 Aug 2024 11:23:55 +0000 (14:23 +0300)]
Fix history accumulating in utils_cache
Eero Tamminen [Tue, 16 Apr 2024 18:05:26 +0000 (21:05 +0300)]
gpu_sysman: remove _total suffix for monotonic metrics
As "write_prometheus" plugin already adds that unconditionally to
names of all monotonic metric types.
Signed-off-by: Eero Tamminen <eero.t.tamminen@intel.com>
Eero Tamminen [Tue, 16 Apr 2024 15:54:10 +0000 (18:54 +0300)]
gpu_sysman: usecs/ujoules -> seconds/joules metric types
With the new floating point counter type, joules & seconds can be used
as metric units instead of microjoules & microseconds.
(Frequencies are are still left as Mhz, but for convenience, and
compability with Intel XPU Manager.)
Signed-off-by: Eero Tamminen <eero.t.tamminen@intel.com>
Eero Tamminen [Tue, 16 Apr 2024 15:49:53 +0000 (18:49 +0300)]
gpu_sysman: use new up/down types for cumulative metrics
Signed-off-by: Eero Tamminen <eero.t.tamminen@intel.com>
Eero Tamminen [Fri, 23 Feb 2024 14:54:13 +0000 (16:54 +0200)]
gpu_sysman: use counter_diff() to handle counter overflow
Assumes overflowing counters to be either at 32-bit or 64-bit,
which should cover most common cases, as long as overflow happens
at max once during query period.
Signed-off-by: Eero Tamminen <eero.t.tamminen@intel.com>
Eero Tamminen [Wed, 21 Feb 2024 14:02:34 +0000 (16:02 +0200)]
gpu_sysman: fix ras_submit() argument type
Reported-By: Florian Foster
Signed-off-by: Eero Tamminen <eero.t.tamminen@intel.com>
Eero Tamminen [Wed, 27 Mar 2024 09:57:23 +0000 (11:57 +0200)]
Merge pull request #4293 from eero-t/xe-kmd-support
[collectd 6] gpu_sysman: switch from zeInit() to zesInit()
Eero Tamminen [Thu, 14 Mar 2024 14:39:52 +0000 (16:39 +0200)]
gpu_sysman: use glob.h/libgen.h only if they are present
Signed-off-by: Eero Tamminen <eero.t.tamminen@intel.com>
Eero Tamminen [Mon, 11 Mar 2024 18:49:25 +0000 (20:49 +0200)]
gpu_sysman: add "dev_name" + get "pci_dev" label value from sysfs
Signed-off-by: Eero Tamminen <eero.t.tamminen@intel.com>
Eero Tamminen [Fri, 1 Mar 2024 18:16:25 +0000 (20:16 +0200)]
gpu_sysman: set legacy env var for older backends
Signed-off-by: Eero Tamminen <eero.t.tamminen@intel.com>
Eero Tamminen [Fri, 1 Mar 2024 12:52:49 +0000 (14:52 +0200)]
configure.ac: update gpu_sysman plugin checks for zesInit()
Signed-off-by: Eero Tamminen <eero.t.tamminen@intel.com>
Eero Tamminen [Wed, 28 Feb 2024 17:29:50 +0000 (19:29 +0200)]
gpu_sysman: minor tweaks for GPU info output
Signed-off-by: Eero Tamminen <eero.t.tamminen@intel.com>
Eero Tamminen [Wed, 28 Feb 2024 14:02:19 +0000 (16:02 +0200)]
gpu_sysman: drop L0 core items not supported with zesInit()
When just Sysman is initialized (by using zesInit() instead of
zeInit()), L0 core functions are no-ops and core structs are not
filled (except for UUID), so drop those.
Try also to present the datata in slightly more readable format
by re-ordering it a bit and making indentation more consistent.
As Sysman does not provide device PCI ID for "pci_dev" label,
take device (model/marketing) name instead, and assing it to
"dev_name" label (like Intel XPU Manager already does).
Changing the "pci_dev" member name, is left for later, as it
would conflict with resource attribute and OpenTelemetry metric
name work being done in parallel.
Signed-off-by: Eero Tamminen <eero.t.tamminen@intel.com>
Eero Tamminen [Tue, 27 Feb 2024 12:30:51 +0000 (14:30 +0200)]
gpu_sysman: show also mem health + allocatable amount at init
Sysman may return zero for physical memory size, but allocatable size
should always be valid. This can be useful in identifying different
GPU types (on same machine) from each other.
Skip showing of invalid memory information.
Signed-off-by: Eero Tamminen <eero.t.tamminen@intel.com>
Eero Tamminen [Mon, 26 Feb 2024 20:05:32 +0000 (22:05 +0200)]
gpu_sysman: change from L0 core to Sysman memory APIs
In the GPU info logging function, as L0 core functions do nothing when
(first) init call (zesInit) is done for Sysman.
Signed-off-by: Eero Tamminen <eero.t.tamminen@intel.com>
Eero Tamminen [Fri, 23 Feb 2024 16:57:25 +0000 (18:57 +0200)]
gpu_sysman: zeInit() -> zesInit(), to support Xe GPU KMD
zesInit() requires Level-Zero frontend v1.9 implementing spec v1.5.
Intel L0 Sysman backend supports only "i915" GPU KMD uAPI variants
when initialized using L0 core zeInit() function. Support for the new
"xe" GPU KMD uAPI requires backend to be initialized using (new)
Sysman-specific zesInit() function instead.
Use of zesInit() means that called driver and device functions also
need to be be L0 Sysman versions instead L0 core versions (skipping L0
core initialization may speed plugin initialization).
Signed-off-by: Eero Tamminen <eero.t.tamminen@intel.com>
Eero Tamminen [Thu, 14 Mar 2024 15:34:43 +0000 (17:34 +0200)]
Merge pull request #4177 from eero-t/log-metrics-option
[Collectd 6] gpu_sysman: add "LogMetrics" option
Eero Tamminen [Wed, 28 Feb 2024 14:01:41 +0000 (16:01 +0200)]
gpu_sysman: WA for local & CI clang-format conflict
Comment prevents local clang-format from trying to re-order
includes differently from the clang-format in CI.
Signed-off-by: Eero Tamminen <eero.t.tamminen@intel.com>
Eero Tamminen [Fri, 23 Feb 2024 13:49:06 +0000 (15:49 +0200)]
gpu_sysman: support logging all (new) metric types
New types are not yet used by the plugin, but soon are.
Signed-off-by: Eero Tamminen <eero.t.tamminen@intel.com>
Eero Tamminen [Fri, 23 Feb 2024 13:32:33 +0000 (15:32 +0200)]
gpu_sysman: skip common metric prefix in metrics logging
Signed-off-by: Eero Tamminen <eero.t.tamminen@intel.com>
Eero Tamminen [Wed, 29 Nov 2023 14:59:12 +0000 (16:59 +0200)]
gpu_sysman: document "LogMetrics" option
Signed-off-by: Eero Tamminen <eero.t.tamminen@intel.com>
Eero Tamminen [Fri, 2 Jun 2023 15:55:20 +0000 (18:55 +0300)]
gpu_sysman: Add "LogMetrics" option
This can be used for local real-time monitoring of GPU metrics without
need to pull/parse collectd write plugin output.
Output is most readable when only one metric type and MetricsOutput
variant are enabled, and collectd sees only single GPU. Alternatively
one could grep the output for the relevant GPU, metric type and its
output variant.
Signed-off-by: Eero Tamminen <eero.t.tamminen@intel.com>
Florian Forster [Mon, 26 Feb 2024 20:21:10 +0000 (21:21 +0100)]
Merge pull request #4288 from octo/6/release
ci: Trigger a website update after uploading a release artifact.
Florian Forster [Mon, 26 Feb 2024 16:00:15 +0000 (17:00 +0100)]
Merge pull request #4273 from octo/6/processes
[collectd 6] processes plugin: Migrate to collectd 6 data structures.
Florian Forster [Sat, 24 Feb 2024 22:28:51 +0000 (23:28 +0100)]
processes plugin: Convert appropriate metrics to "up_down".
Florian Forster [Sat, 24 Feb 2024 22:24:12 +0000 (23:24 +0100)]
processes plugin: Rename "fpcounter" to "counter_fp".
Florian Forster [Sat, 24 Feb 2024 22:13:47 +0000 (23:13 +0100)]
processes plugin: Fix reference of a non-existent variable.
Florian Forster [Sat, 24 Feb 2024 22:10:54 +0000 (23:10 +0100)]
processes plugin: Consistently call `procstat_t*` variables "ps".
Florian Forster [Sat, 24 Feb 2024 22:07:51 +0000 (23:07 +0100)]
processes plugin: Unify the usage of `read_text_file_contents()`.
Florian Forster [Sat, 24 Feb 2024 21:52:40 +0000 (22:52 +0100)]
processes plugin: Add a comment to point readers at more documentation.
Florian Forster [Sat, 24 Feb 2024 21:45:45 +0000 (22:45 +0100)]
processes plugin: Remove obsolete code.
Florian Forster [Sat, 24 Feb 2024 21:38:07 +0000 (22:38 +0100)]
processes plugin: Fix incorrect scaling of `process.cpu.utilization`.
Florian Forster [Sat, 24 Feb 2024 21:32:10 +0000 (22:32 +0100)]
processes plugin: Remove unnecessarily verbose assertion.
Florian Forster [Sat, 24 Feb 2024 21:31:52 +0000 (22:31 +0100)]
processes plugin: Return boolean from `process_resource()`.
Florian Forster [Sat, 24 Feb 2024 21:31:20 +0000 (22:31 +0100)]
processes plugin: Remove unused struct field.
Florian Forster [Tue, 6 Feb 2024 14:27:14 +0000 (15:27 +0100)]
processes plugin: Fix assignment of CPU time counters.
Florian Forster [Tue, 6 Feb 2024 09:44:03 +0000 (10:44 +0100)]
processes plugin: Remove unused fields from `struct procstat`.
Florian Forster [Tue, 6 Feb 2024 09:39:16 +0000 (10:39 +0100)]
processes plugin: Report delay accounting in seconds.
Florian Forster [Tue, 6 Feb 2024 09:27:23 +0000 (10:27 +0100)]
processes plugin: Report context switches as metric family.
Florian Forster [Tue, 6 Feb 2024 09:23:13 +0000 (10:23 +0100)]
processes plugin: Report the fork rate as metric family.
Florian Forster [Tue, 6 Feb 2024 08:47:37 +0000 (09:47 +0100)]
processes plugin: Report process CPU time in seconds.
Florian Forster [Fri, 19 Jan 2024 19:55:10 +0000 (20:55 +0100)]
processes plugin: Port CPU usage to `metric_family_t`.
Florian Forster [Fri, 19 Jan 2024 08:36:53 +0000 (09:36 +0100)]
processes plugin: Use `sysconf(_SC_CLK_TCK)` instead of `CONFIG_HZ`.
Florian Forster [Fri, 19 Jan 2024 08:27:13 +0000 (09:27 +0100)]
processes plugin: Add note about AIX using `struct timeval` weirdly.
Florian Forster [Thu, 18 Jan 2024 16:03:37 +0000 (17:03 +0100)]
processes plugin: Port the delay accounting to `metric_family_t`.
Florian Forster [Wed, 17 Jan 2024 15:57:18 +0000 (16:57 +0100)]
processes plugin: Port metrics to new data structure:
* I/O bytes
* I/O operations
* number of file descriptors
* number of page faults
Florian Forster [Wed, 17 Jan 2024 13:26:15 +0000 (14:26 +0100)]
processes plugin: Initial working code for collectd 6.
Florian Forster [Tue, 16 Jan 2024 06:47:40 +0000 (07:47 +0100)]
configure.ac: Enable the processes plugin by default.
Florian Forster [Mon, 15 Jan 2024 21:18:41 +0000 (22:18 +0100)]
processes plugin: Convert the aggregate system metric.
Florian Forster [Mon, 15 Jan 2024 20:29:54 +0000 (21:29 +0100)]
processes plugin: Break the enormous `ps_read` function into OS specific parts.
Florian Forster [Sat, 24 Feb 2024 21:25:35 +0000 (22:25 +0100)]
ci: Trigger a website update after uploading a release artifact.
Florian Forster [Wed, 21 Feb 2024 21:08:17 +0000 (22:08 +0100)]
Update ChangeLog for version 6.0.0.rc3.
Florian Forster [Wed, 21 Feb 2024 20:58:11 +0000 (21:58 +0100)]
Merge pull request #4287 from octo/6/up_down_counter
[collectd 6] Implement "UpDownCounter" metric types.
Florian Forster [Wed, 21 Feb 2024 14:47:13 +0000 (15:47 +0100)]
memory plugin: Remove unnecessary casts.
Florian Forster [Wed, 21 Feb 2024 14:41:41 +0000 (15:41 +0100)]
common: Improve error case description.
Florian Forster [Wed, 21 Feb 2024 12:38:02 +0000 (13:38 +0100)]
open_telemetry plugin: Receiver: Add support for counter_fp, up_down, and up_down_fp.
Florian Forster [Wed, 21 Feb 2024 08:56:40 +0000 (09:56 +0100)]
Tree wide: Remove `fpcounter_t` and use `double` instead.
Florian Forster [Wed, 21 Feb 2024 08:55:04 +0000 (09:55 +0100)]
Tree wide: Remove `up_down_counter_fp_t` and use `double` instead.
Florian Forster [Wed, 21 Feb 2024 08:53:37 +0000 (09:53 +0100)]
Tree wide: Remove `up_down_counter_t` and use `int64_t` instead.
Florian Forster [Wed, 21 Feb 2024 08:51:24 +0000 (09:51 +0100)]
Tree wide: Rename "fpcounter" to "counter_fp".
Florian Forster [Wed, 21 Feb 2024 08:49:23 +0000 (09:49 +0100)]
Tree wide: Rename "up_down_counter_fp" to "up_down_fp".
Florian Forster [Wed, 21 Feb 2024 08:46:57 +0000 (09:46 +0100)]
Tree wide: Rename "up_down_counter" to "up_down".
Florian Forster [Wed, 21 Feb 2024 08:36:43 +0000 (09:36 +0100)]
contrib/format.sh <all the things>
Florian Forster [Fri, 9 Feb 2024 10:04:29 +0000 (11:04 +0100)]
df plugin: Migrate to "up down counter".
Florian Forster [Fri, 9 Feb 2024 09:57:22 +0000 (10:57 +0100)]
memory plugin: Migrate to "up down counter".
Florian Forster [Fri, 9 Feb 2024 07:55:30 +0000 (08:55 +0100)]
Daemon: add "up down counters" in both an integer and floating point variant.
Florian Forster [Wed, 21 Feb 2024 20:05:40 +0000 (21:05 +0100)]
Merge pull request #4284 from octo/6/write_prometheus
[collectd 6] write_prometheus plugin: Fix handling of multiple resources.
Florian Forster [Wed, 21 Feb 2024 16:13:56 +0000 (17:13 +0100)]
write_prometheus plugin: Unconditionally print the job and instance labels.
Florian Forster [Wed, 21 Feb 2024 16:11:44 +0000 (17:11 +0100)]
write_prometheus plugin: Update unit tests to the new semantics.
Florian Forster [Wed, 21 Feb 2024 15:20:30 +0000 (16:20 +0100)]
write_prometheus plugin: Let "job" default to hostname_g, and "instance" to an empty string.
Florian Forster [Tue, 20 Feb 2024 17:21:14 +0000 (18:21 +0100)]
contrib/format.sh src/write_prometheus.c
Florian Forster [Tue, 20 Feb 2024 17:19:05 +0000 (18:19 +0100)]
write_prometheus plugin: Fix updating of existing metrics (+test).
Eero Tamminen [Tue, 20 Feb 2024 15:17:06 +0000 (17:17 +0200)]
Merge pull request #4223 from eero-t/sysman-test-metrics
[collectd 6] Use real collectd metric functions in gpu_sysman plugin tests
Florian Forster [Tue, 20 Feb 2024 14:53:45 +0000 (15:53 +0100)]
write_prometheus plugin: Fix formatting in the unit test.
Florian Forster [Tue, 20 Feb 2024 14:53:26 +0000 (15:53 +0100)]
write_prometheus plugin: Simplify files included by the test.
Florian Forster [Tue, 20 Feb 2024 14:50:17 +0000 (15:50 +0100)]
write_prometheus plugin: Complete error handling in `prom_metric_family_clone()`.
Florian Forster [Tue, 20 Feb 2024 14:46:48 +0000 (15:46 +0100)]
write_prometheus plugin: Make all functions `static` again.
Florian Forster [Tue, 20 Feb 2024 07:23:38 +0000 (08:23 +0100)]
Revert "Daemon: Add `metric_list_append`."
This reverts commit
0b3a481b457038b048c19df2af656a925644273f .
Florian Forster [Mon, 19 Feb 2024 22:22:34 +0000 (23:22 +0100)]
contrib/format.sh src/write_prometheus.c src/write_prometheus_test.c
Florian Forster [Mon, 19 Feb 2024 22:17:41 +0000 (23:17 +0100)]
write_prometheus plugin: Refactor the internal data structures.
(Ab)Using the `metric_t` and `metric_family_t` data structures provided by the
daemon made memory management quite hard and therefore brittle.
This introduces types specific to the *write_prometheus plugin* that store the
resource attributes with the metric, not the family.
Florian Forster [Mon, 19 Feb 2024 20:56:02 +0000 (21:56 +0100)]
write_prometheus plugin: Store resource attributes with the metric, not the metric family.
For Prometheus output, the plugin groups all metrics with the same name into
one `metric_family_t`. This caused problems when collectd handled metrics from
multiple resources.
To solve this issue, we're somewhat abusing the data structure and store
per-metric resource attributes in the `family` field. That means for the
metrics stored in the *write_prometheus plugin* `(metric_t).family` does not
point back to the metric family containing the metric.
Fixes: #4283
Florian Forster [Mon, 19 Feb 2024 20:54:45 +0000 (21:54 +0100)]
Daemon: Add `metric_list_append`.
Florian Forster [Mon, 19 Feb 2024 17:24:35 +0000 (18:24 +0100)]
write_prometheus plugin: Add an end-to-end test.
Florian Forster [Tue, 20 Feb 2024 14:40:18 +0000 (15:40 +0100)]
Merge pull request #4271 from octo/6/otelcol
[collectd 6] New plugin (sort of): OpenTelemetry receiver
Florian Forster [Tue, 20 Feb 2024 14:17:53 +0000 (15:17 +0100)]
open_telemetry plugin: Move prototypes to a header file.
Florian Forster [Sun, 4 Feb 2024 07:46:23 +0000 (08:46 +0100)]
open_telemetry: Unify the configuration handling between exporter and receiver.
Florian Forster [Sun, 4 Feb 2024 07:45:11 +0000 (08:45 +0100)]
open_telemetry plugin: Minor cleanups.
* Set field to `NULL` after freeing.
* Remove unused global variable.
Florian Forster [Sun, 4 Feb 2024 07:43:33 +0000 (08:43 +0100)]
open_telemetry plugin: Fix spelling in the manpage.
Florian Forster [Fri, 2 Feb 2024 16:05:08 +0000 (17:05 +0100)]
open_telemetry plugin: Add support for floating point counters.
Florian Forster [Fri, 2 Feb 2024 15:18:19 +0000 (16:18 +0100)]
open_telemetry plugin: Update the synopsis in the manpage.
Florian Forster [Fri, 2 Feb 2024 15:15:51 +0000 (16:15 +0100)]
open_telemetry plugin: Add SSL options for the exporter.
Florian Forster [Fri, 2 Feb 2024 12:51:15 +0000 (13:51 +0100)]
open_telemetry plugin: Unify the configuration between exporter and receiver.
Florian Forster [Fri, 2 Feb 2024 12:31:37 +0000 (13:31 +0100)]
open_telemetry plugin: Unify the prefix used in log messages.
Florian Forster [Fri, 2 Feb 2024 12:21:17 +0000 (13:21 +0100)]
open_telemetry plugin: Combine the *open_telemetry_collector* and *write_open_telemetry* plugins.
Florian Forster [Thu, 1 Feb 2024 11:53:31 +0000 (12:53 +0100)]
open_telemetry_collector plugin: Enable gRPC reflection if available.
Florian Forster [Thu, 1 Feb 2024 11:36:01 +0000 (12:36 +0100)]
open_telemetry_collector plugin: Populate `ssl_opts` only when needed.
Florian Forster [Thu, 1 Feb 2024 11:34:15 +0000 (12:34 +0100)]
open_telemetry_collector plugin: Link with protobuf library.