mirror of
https://github.com/prometheus/node_exporter.git
synced 2024-11-09 23:24:09 -08:00
Make filesystem space prediction window configurable (#2844)
Signed-off-by: fitz123 <alugovoi@ordercapital.com>
This commit is contained in:
parent
7333465abf
commit
81fc05c45f
|
@ -10,7 +10,7 @@
|
||||||
(
|
(
|
||||||
node_filesystem_avail_bytes{%(nodeExporterSelector)s,%(fsSelector)s,%(fsMountpointSelector)s} / node_filesystem_size_bytes{%(nodeExporterSelector)s,%(fsSelector)s,%(fsMountpointSelector)s} * 100 < %(fsSpaceFillingUpWarningThreshold)d
|
node_filesystem_avail_bytes{%(nodeExporterSelector)s,%(fsSelector)s,%(fsMountpointSelector)s} / node_filesystem_size_bytes{%(nodeExporterSelector)s,%(fsSelector)s,%(fsMountpointSelector)s} * 100 < %(fsSpaceFillingUpWarningThreshold)d
|
||||||
and
|
and
|
||||||
predict_linear(node_filesystem_avail_bytes{%(nodeExporterSelector)s,%(fsSelector)s,%(fsMountpointSelector)s}[6h], 24*60*60) < 0
|
predict_linear(node_filesystem_avail_bytes{%(nodeExporterSelector)s,%(fsSelector)s,%(fsMountpointSelector)s}[%(fsSpaceFillingUpPredictionWindow)s], 24*60*60) < 0
|
||||||
and
|
and
|
||||||
node_filesystem_readonly{%(nodeExporterSelector)s,%(fsSelector)s,%(fsMountpointSelector)s} == 0
|
node_filesystem_readonly{%(nodeExporterSelector)s,%(fsSelector)s,%(fsMountpointSelector)s} == 0
|
||||||
)
|
)
|
||||||
|
|
|
@ -54,13 +54,19 @@
|
||||||
// 'NodeFilesystemSpaceFillingUp' alerts. These alerts fire if the disk
|
// 'NodeFilesystemSpaceFillingUp' alerts. These alerts fire if the disk
|
||||||
// usage grows in a way that it is predicted to run out in 4h or 1d
|
// usage grows in a way that it is predicted to run out in 4h or 1d
|
||||||
// and if the provided thresholds have been reached right now.
|
// and if the provided thresholds have been reached right now.
|
||||||
// In some cases you'll want to adjust these, e.g. by default Kubernetes
|
// In some cases you'll want to adjust these, e.g., by default, Kubernetes
|
||||||
// runs the image garbage collection when the disk usage reaches 85%
|
// runs the image garbage collection when the disk usage reaches 85%
|
||||||
// of its available space. In that case, you'll want to reduce the
|
// of its available space. In that case, you'll want to reduce the
|
||||||
// critical threshold below to something like 14 or 15, otherwise
|
// critical threshold below to something like 14 or 15, otherwise
|
||||||
// the alert could fire under normal node usage.
|
// the alert could fire under normal node usage.
|
||||||
|
// Additionally, the prediction window for the alert can be configured
|
||||||
|
// to account for environments where disk usage can fluctuate within
|
||||||
|
// a short time frame. By extending the prediction window, you can
|
||||||
|
// reduce false positives caused by temporary spikes, providing a
|
||||||
|
// more accurate prediction of disk space issues.
|
||||||
fsSpaceFillingUpWarningThreshold: 40,
|
fsSpaceFillingUpWarningThreshold: 40,
|
||||||
fsSpaceFillingUpCriticalThreshold: 20,
|
fsSpaceFillingUpCriticalThreshold: 20,
|
||||||
|
fsSpaceFillingUpPredictionWindow: '6h',
|
||||||
|
|
||||||
// Available disk space (%) thresholds on which to trigger the
|
// Available disk space (%) thresholds on which to trigger the
|
||||||
// 'NodeFilesystemAlmostOutOfSpace' alerts.
|
// 'NodeFilesystemAlmostOutOfSpace' alerts.
|
||||||
|
|
Loading…
Reference in a new issue