github/node-problem-detector
1
0
Fork 0
mirror of https://github.com/kubernetes/node-problem-detector.git synced 2026-08-19 20:26:32 +00:00
Code Issues Packages Projects Releases Wiki Activity
646 Commits 12 Branches 59 Tags
68bf26b08f68853cb76e5b7a4c89eb12409c35e2
Commit Graph
35 Commits
Author SHA1 Message Date
varsha teratipally 7b51a90328 Reduce the number of reads to /proc/partitions file 
to retrive the partitions on disk
 2021-06-13 21:11:34 +00:00
Kubernetes Prow RobotandGitHub 9c541692ee Merge pull request #557 from vteratipally/adfad 
Make sure the path to known-modules.json is relative
 2021-05-14 14:39:59 -07:00
Varsha Teratipally a79b87ce7e Make sure the path to known-modules.json is relative to the 
system-stats-monitor.json file
 2021-05-14 21:14:55 +00:00
Jeremy Edwards d4933875ed Add support for basic system metrics for Windows. 2021-05-10 21:58:38 +00:00
Karan Goel c2aceee61d remove os_versions and kernel_version labels 2021-02-02 08:25:10 -08:00
Karan Goel 8648fe265a add metric for per-cpu, per-stage timing 2021-01-29 08:46:39 -08:00
Karan Goel 2a2bab3d28 Add network interface stats 
We do not have to collect these often, so for now set the collection
interval to 120s (even though the Stackdriver exporter is still set to
export every 60s).
 2021-01-20 08:56:34 -08:00
Kubernetes Prow RobotandGitHub c2d7a7be62 Merge pull request #513 from karan/cpu_activity_metrics 
add metrics for process stats
 2021-01-19 18:38:07 -08:00
varsha teratipally 2cb1195f18 cleanup the log 2021-01-13 17:54:53 +00:00
Karan Goel f13d2a5449 don't run os feature collector if metric not initialized 2021-01-13 09:33:13 -08:00
Karan Goel 71098097c0 add metrics for process stats 
Tested on a COS VM:

```
$ curl -s localhost:20257/metrics | grep "^system_"
system_interrupts_total{kernel_version="5.4.49+",os_version="cos 85-13310.1041.24"} 8.759236e+07
system_processes_total{kernel_version="5.4.49+",os_version="cos 85-13310.1041.24"} 692506
system_procs_blocked{kernel_version="5.4.49+",os_version="cos 85-13310.1041.24"} 0
system_procs_running{kernel_version="5.4.49+",os_version="cos 85-13310.1041.24"} 2
```
 2021-01-13 09:14:08 -08:00
varsha teratipally f89f620909 added new line in the known_modules.json 2021-01-08 23:25:02 +00:00
varsha teratipally eb38b4b598 added a new metric to retrieve os features like unknown modules 2021-01-08 21:52:16 +00:00
varsha teratipally 047958a49c changing the label names as per the standards 2020-12-08 02:27:22 +00:00
Kubernetes Prow RobotandGitHub 112d53b10a Merge pull request #497 from vteratipally/fs_types 
avoid duplicating the disk bytes used metrics based on fstype and mount types
 2020-11-18 10:48:07 -08:00
Kubernetes Prow RobotandGitHub 33571a312d Merge pull request #478 from neoseele/master 
fix: node memory metrics are off by 1024
 2020-11-16 14:06:12 -08:00
varsha teratipally 1550882948 avoid duplicating the disk bytes used metrics based on fstype and mountopts 2020-11-16 20:10:46 +00:00
Karan Goel 925ea7393c Collect CPU load averages in a separate metric 2020-11-09 09:41:52 -08:00
Neil 589411702a fix: node memory metrics are off by 1024 
The memory unit in /proc/meminfo is kB (b/171164235)

```
MemTotal:       264129908 kB
MemFree:        153559480 kB
...
```
 2020-10-19 17:26:31 +11:00
varsha teratipally 50127b0512 changed labelname after code review 2020-08-06 00:43:45 +00:00
varsha teratipally 4c40b7e468 updated readme 2020-08-05 21:43:58 +00:00
varsha teratipally e13210157d Add more info to disk metrics 2020-08-05 21:12:25 +00:00
Xuewei Zhang 83b09277f0 Collect more cpu/disk/memory metrics 2020-02-03 15:29:45 -08:00
Xuewei Zhang fa7a3d7df1 Fix disk metrics unit and queue_length calculation 2020-01-02 17:19:38 -08:00
Alex Wong 5a4ac81186 Only disk_avg_queue_len is distorted on first collection 2019-12-12 14:39:29 +08:00
Alex Wong 3d10c892a2 Ignore first collected disk stats to prevent metric distortion 2019-12-11 11:14:01 +08:00
Xuewei Zhang 9e789b5f99 Refactor on metrics so that names for all the views are tracked 2019-09-11 12:07:13 -07:00
Xuewei Zhang 0f2fce56e5 Change host/uptime to GAUGE metrics 2019-09-10 16:58:06 -07:00
Xuewei Zhang 82c2368795 Metric format fixes on host/uptime and disk/* 
1. host/uptime, disk/io_time and disk/weighted_io should be
counter/cumulative metrics. SO we have to use the Sum aggregation method
rather than LastValue aggregation method (which will declare the metric
as gauge metric).

2. Renamed label "device" for disk/* metrics to "device_name".
This is to clarify that it is device_name (sda1) rather than device_path
(/dev/sda1)
 2019-08-16 15:14:54 -07:00
Zhen Wang 182a9450dd Print monitor config path in the logs 2019-07-30 11:00:47 -07:00
Xuewei Zhang fbebcf311b Report metrics from system-log-monitor 2019-07-12 14:38:21 -07:00
Xuewei Zhang 4944ac3e48 Implement host collector as part of system-stats-monitor 
Host collector report three things today:
1. Host OS uptime (in seconds)
2. Host kernel version (as a metric label)
3. Host OS version (as a metric label)
 2019-06-27 16:40:11 -07:00
Xuewei Zhang 29b0740f4c Refactor systemstatsmonitor/metric_helper.go into a metrics package 2019-06-27 16:40:05 -07:00
Xuewei Zhang cf6624661a Update READMEs 2019-06-13 00:51:17 -07:00
Xuewei Zhang 7ad5dec712 Add disk metrics support. 2019-06-13 00:51:17 -07:00