Skip to main content

Home

Resource metrics

The table below lists the resource metrics collected by Unravel. The availability of a particular metric is dependent on the underlying OS and JVM GC algorithm in use.

Metric

Unit

Description

allocatedBytes

bytes

Accumulated number of allocated bytes.

availableMemory

bytes

An estimate of memory available for launching new processes.

avgFullGcInterval

nanoseconds (duration)

Average interval between two subsequent full GCs. Might not be available for particular GC algorithms.

avgMinorInterval

nanoseconds (duration)

Average interval between two subsequent minor GCs. Might not be available for particular GC algorithms.

blockingRatio

percent

Estimated percentage of CPU time spent in kernel blocking operations.

committedHeap

bytes

The committed heap size.

committedNonHeap

bytes

The committed non-heap size.

committedVirtualMemory

bytes

The committed virtual memory in the operating system.

currentThreadCpuTime

nanoseconds (duration)

Current thread CPU time elapsed since the start of the measurement. Might not be available on some operating systems.

currentThreadUserTime

nanoseconds (duration)

Current thread user time elapsed since the start of the measurement. Might not be available on some operating systems.

edenPeakUsage

bytes

Maximum memory usage in the eden space.

freePhysicalMemory

bytes

The free physical memory in the operating system.

freeSwap

bytes

The free swap size.

fullGcCount

count

Number of full GC runs.

fullGcTime

nanoseconds (duration)

Accumulated time spent in full GC.

gcEdenSurvivedAvg

bytes

Average number of bytes moved from eden to survivor space. Might not be available for particular GC algorithms.

gcLoad

percent

Percentage of CPU time spent in GC.

gcOldLiveAvg

bytes

Average number of bytes alive in the old generation. Might not be available for particular GC algorithms.

gcSurvivorPromotedAvg

bytes

Average number of bytes moved from survivor to old space. Might not be available for particular GC algorithms.

gcYoungLiveAvg

bytes

Average number of bytes alive in the young generation (eden + survivor spaces). It might not be available for particular GC algorithms.

initHeap

bytes

The initial heap size.

initNonHeap

bytes

The initial non-heap size.

maxHeap

bytes

The maximum heap size.

maxNonHeap

bytes

The maximum non-heap size.

minorGcCount

COUNT

The number of minor GC runs.

minorGcTime

nanoseconds (duration)

The accumulated time spent in minor GC.

oldPeakUsage

bytes

The maximum memory usage in the old space.

processCpuLoad

PERCENT

Average process CPU load for the last minute (all cores).

snapshotTs

milliseconds (timestamp)

The time the metric was read.

startTs

milliseconds (timestamp)

The time when the collection process started.

survivorPeakUsage

bytes

Maximum memory usage in the survivor space.

systemCpuLoad

PERCENT

Average system CPU load for the last minute (all cores).

totalPhysicalMemory

bytes

The total physical memory in the operating system.

totalSwap

bytes

The total swap size.

usedHeap

bytes

The used heap size.

usedNonHeap

bytes

The used non-heap size.

vmRss

bytes

The resident set size of the complete process tree.

vmRssDir

bytes

The resident set size of the process.

Notice

he vmRss metric values can underestimate the memory footprint of objects that are not directly managed by the JVM. More accurate metric values can be obtained by adding the property -Dcom.unraveldata.metrics.proctree.enable=true to spark.driver.extraJavaOptions and spark.executor.extraJavaOptions