Update system monitor.

Overall cleanup, and modificaton to the disk chart.  System tracks Byte,
KiB and MiB transfer rates.  Can't realld support GiB because I'm no
sure the PCIe bus can handle that.

Modifed the Y-Axis for disks so it displays the data on the closest
power of 2 scale. Seems to be a reasonable way to modify the scale so we
can actually see some of the smaller writes.
This commit is contained in:
Jeff Curless
2025-10-13 17:59:47 -04:00
parent 3f0727e590
commit cfadd92430
4 changed files with 529 additions and 54 deletions

View File

@@ -3,13 +3,38 @@ import os
import time
class CPULoad:
'''
A class to help with obtaining the CPU load of the system. If there is more information
needed, we can add to this.
Note:
This code automatically attempts to load the data from the system to initialize the
object with names, and an initial set of data.
This may result in th first actual call return some not very consistent values, for
the time period being observed, but that difference is minimal. In otherwords if we
the period of time being measured is 1 second, and it's been a minute since this class
was initialized, the first period reported will be CPU load over the minute, not 1 second,
and the second period reported will be for a second...
This is usually not an issue.
'''
def __init__( self ):
self._previousData = self._getRawData()
self._names = []
self._previousData : dict[str,tuple] = self._getRawData()
self._names : list[str] = []
for item in self._previousData:
self._names.append( item )
def _getRawData( self ):
def _getRawData( self ) -> dict[str : tuple]:
'''
Obtain the raw CPU data from the system (located in /prop/stat), and
return just the cpu0 -> cpux values. No assumption is made on the number of
cpus.
Returns:
A dictionary is returned, the format is name = (total, idle). The total
time and idle time are use to determine the percent utilization of the system.
'''
result = {}
with open( "/proc/stat", "r") as f:
allLines = f.readlines()
@@ -25,22 +50,50 @@ class CPULoad:
result[cpu[0]] = (total,idle)
return result
def getPercentages( self ):
def getPercentages( self ) -> dict[ str : float ]:
'''
Obtain the percent CPU utilization of the system for a period of time.
This routine gets the current raw data from the system, and then performs
a delta from the prior time this function was called. This data is then run
through the following equation:
utilization = ((total - idle)/total) * 100
If the snapshots are taken at relativy consistent intervals, the CPU
utilization in percent, is reasonably lose to the actual percentage.
Returns:
A dictionary consisting of the name of the CPU, and a floating point
number representing the current utilization of that CPU.
'''
results = {}
current = self._getRawData()
for item in current:
total = current[item][0] - self._previousData[item][0]
idle = current[item][1] - self._previousData[item][1]
percent = ((total - idle)/total) * 100
results[item] = percent
results[item] = round(percent,2)
self._previousData = current
return results
@property
def cpuNames( self ):
def cpuNames( self ) -> list[str]:
'''
Get a list of CPU names from the system.
Returns:
a list of strings
'''
return self._names
def __len__(self):
def __len__(self) -> int:
'''
handle getting the length (or count of CPU's).
Returns:
Number of CPU's
'''
return len(self._previousData)
if __name__ == "__main__":
@@ -48,6 +101,7 @@ if __name__ == "__main__":
print( f"Number of CPU's = {len(load)}" )
while True:
time.sleep( 1 )
percentage = load.getPercentages()
percentage : dict[str:float] = load.getPercentages()
print( f"percentage: {percentage}" )
for item in percentage:
print( f"{item} : {percentage[item]:.02f}" )

View File

@@ -7,12 +7,9 @@ Requires: PyQt5 (including QtCharts)
"""
import sys
from typing import Tuple, List
from gpiozero import CPUTemperature
from oneUpSupport import systemData
from cpuload import CPULoad
import os
import sys
from systemsupport import systemData, CPULoad
# --------------------------
# Globals
@@ -30,25 +27,27 @@ from PyQt5.QtWidgets import QApplication, QMainWindow, QWidget, QGridLayout, QLa
from PyQt5.QtChart import QChart, QChartView, QLineSeries, QValueAxis
class RollingChart(QWidget):
"""
'''
A reusable chart widget with one or more QLineSeries and a rolling X window.
Args:
title: Chart title
series_defs: List of (name, color_qt_str or None) for each line
y_min, y_max: Fixed Y axis range
window: number of points to keep (points are 1 per tick by default)
"""
def __init__(self, title: str, series_defs: List[tuple], y_min: float, y_max: float, window: int = 120, parent=None):
Parameters:
title - Chart title.
series_defs - List of (name, color_qt_str or None) for each line.
y_min,y_max - Fixed Y axis range.
window - Number of points to keep (points are 1 per tick by default).
'''
def __init__(self, title: str, series_defs: list[tuple], y_min: float, y_max: float, window: int = 120, parent=None):
super().__init__(parent)
self.window = window
self.x = 0
self.series: List[QLineSeries] = []
self.xpos = window - 1
self.chart = QChart()
self.chart.setTitle(title)
self.chart.legend().setVisible(len(series_defs) > 1)
self.chart.legend().setAlignment(Qt.AlignBottom)
self.series:list[QLineSeries] = []
for name, color in series_defs:
s = QLineSeries()
s.setName(name)
@@ -57,14 +56,20 @@ class RollingChart(QWidget):
self.series.append(s)
self.chart.addSeries(s)
# Axes
# Setup X Axis... Note, setVisible disables all of this, however whatI
# want is the tick count etc, but NO lable on the axis. There does not
# appear to be a way to do that.
self.axis_x = QValueAxis()
self.axis_x.setRange(0, self.window)
#self.axis_x.setTitleText("Seconds")
self.axis_x.setMinorTickCount( 2 )
self.axis_x.setTickCount( 10 )
self.axis_x.setLabelFormat("%d")
self.axis_x.setVisible(False)
# Setup Y Axis...
self.axis_y = QValueAxis()
self.axis_y.setRange(y_min, y_max)
self.axis_y.setLabelFormat( "%d" )
self.chart.addAxis(self.axis_x, Qt.AlignBottom)
self.chart.addAxis(self.axis_y, Qt.AlignLeft)
@@ -80,26 +85,31 @@ class RollingChart(QWidget):
layout.setContentsMargins(0, 0, 0, 0)
layout.addWidget(self.view, 0, 0)
def append(self, values: List[float]):
"""
def append(self, values: list[float]):
'''
Append one sample (for each series) at the next x value. Handles rolling window.
values must match the number of series.
"""
self.x += 1
Parameters:
values - A list of floating point numbers, on per data series in the
chart.
'''
self.xpos += 1
for s, v in zip(self.series, values):
# Handle NaN by skipping, or plot zero—here we clamp None/NaN to None and skip
try:
if v is None:
continue
# If you want to clamp, do it here: v = max(self.axis_y.min(), min(self.axis_y.max(), v))
s.append(self.x, float(v))
except Exception:
s.append(self.xpos, float(v))
except Exception as error:
# ignore bad data points
print( f"Exception error {error}" )
pass
# Trim series to rolling window
min_x_to_keep = max(0, self.x - self.window)
self.axis_x.setRange(min_x_to_keep, self.x)
min_x_to_keep = max(0, self.xpos - self.window)
self.axis_x.setRange(min_x_to_keep, self.xpos)
for s in self.series:
# Efficient trim: remove points with x < min_x_to_keep
@@ -116,11 +126,141 @@ class RollingChart(QWidget):
hi = mid
s.replace(points[lo:]) # keep tail only
class scaleValues:
def __init__( self, range_y ):
self.index = 0
self.valueRange = range_y
@property
def scale( self ):
return self.valueRange[self.index][1]
def scaleValue(self, value : float ):
return value / self.scale
def nextScale( self ):
if (self.index + 1) < len(self.valueRange):
self.index += 1
#print( f"Switched scale to {self.valueRange[self.index]}")
def prevScale( self ):
if self.index > 0:
self.index -= 1
#print( f"Switches scale to {self.valueRange[self.index]}")
def scalePointsDown( self, points ):
amount = self.valueRange[self.index][1]
for point in points:
point.setY(point.y() / amount)
def scaleDown( self, value ):
return value / 1024
def scaleUp( self, value ):
return value * 1024
def scalePointsUp( self, points ):
amount = self.valueRange[self.index][1]
for point in points:
point.setY(point.y() * amount)
@property
def name( self ):
return self.valueRange[self.index][0]
class RollingChartDynamic(RollingChart):
def __init__(self, title : str, series_defs: list[tuple], range_y : list[tuple], window=120,parent=None):
self.maxY = 512
super().__init__(title,series_defs,0,self.maxY,window,parent)
self.title = title
self.max = 0
self.scale = scaleValues(range_y)
self.chart.setTitle( title+ f" ({self.scale.name})" )
def getBestFit( self, value ):
values = [4,8,16,32,64,128,256,512,1024]
for i in values:
if value < i:
return i
return 4
def append(self, values: list[float]):
'''
Append one sample (for each series) at the next x value. Handles rolling window.
values must match the number of series.
Parameters:
values - A list of floating point numbers, on per data series in the
chart.
'''
scaleUp = False
self.xpos += 1
for s, v in zip(self.series, values):
# Handle NaN by skipping, or plot zero—here we clamp None/NaN to None and skip
try:
if v is None:
continue
sv = self.scale.scaleValue(v)
if sv > 1024:
scaleUp = True
# If you want to clamp, do it here: v = max(self.axis_y.min(), min(self.axis_y.max(), v))
#if v:
# print( f"value : {v} scaled: {sv} " )
s.append(self.xpos, float(sv))
except Exception as error:
# ignore bad data points
print( f"Exception error {error}" )
pass
# Trim series to rolling window
min_x_to_keep = max(0, self.xpos - self.window)
self.axis_x.setRange(min_x_to_keep, self.xpos)
if scaleUp:
self.scale.nextScale()
self.chart.setTitle(self.title + f" ({self.scale.name})" )
maxV = 0
for s in self.series:
drop = 0
points = s.pointsVector()
for index, point in enumerate(points):
if point.x() < min_x_to_keep:
drop = index
if scaleUp:
point.setY( self.scale.scaleDown(point.y()))
if maxV < point.y():
maxV = point.y()
s.replace( points[drop:] )
if maxV > 1:
self.axis_y.setRange( 0, self.getBestFit(maxV) )
#print( f"maxV left is {maxV}" )
if maxV < 1:
self.scale.prevScale()
self.chart.setTitle( self.title + f" ({self.scale.name})")
for s in self.series:
points = s.pointsVector()
for point in points:
point.setY( self.scale.scaleUp(point.y()))
s.replace(points)
class MonitorWindow(QMainWindow):
'''
Creating a window to monitor various system aspects.
Parameters:
refresh_ms - Time between refreshes of data on screen, in milliseconds, the
default is 1 second.
window - How much data do we want to store in the graph? Each data point
is a data refresh period.
Parent - Owning parent of this window... default is None.
'''
def __init__(self, refresh_ms: int = 1000, window = 120, parent=None):
super().__init__(parent)
self.setWindowTitle("Argon 1UP Monitor")
self.setWindowTitle("System Monitor")
self.setMinimumSize(900, 900)
central = QWidget(self)
@@ -155,14 +295,14 @@ class MonitorWindow(QMainWindow):
window=window
)
self.io_chart = RollingChart(
title="NVMe I/O (MB/s)",
self.io_chart = RollingChartDynamic(
title="Disk I/O",
series_defs=[
("Read MB/s", None),
("Write MB/s", None),
("Read", None),
("Write", None),
],
y_min=0, y_max=1100, # adjust ceiling for your device
window=window
range_y=[("Bytes/s", 1),("KiB/s",1024),("MiB/s", 1024*1024),("GiB/s",1024*1024*1024)],
window=window,
)
# Layout: 2x2 grid (CPU, NVMe on top; IO full width bottom)
@@ -171,30 +311,40 @@ class MonitorWindow(QMainWindow):
grid.addWidget(self.cpu_chart, 2, 0, 1, 1 )
grid.addWidget(self.fan_chart, 2, 1, 1, 1 )
# Get the initial information from the syste
self.refresh_metrics()
# Timer
self.timer = QTimer(self)
self.timer.timeout.connect(self.refresh_metrics)
self.timer.start(refresh_ms)
self.refresh_metrics()
def refresh_metrics(self):
# Gather metrics with safety
'''
This routine is called periodically, as setup in the __init__ functon. Since this
routine calls out to other things, we want to make sure that there is no possible
exception, so everything needs to be wrapped in a handler.
'''
# Obtain the CPU temperature
try:
cpu_c = float(sysdata.CPUTemperature)
except Exception:
cpu_c = None
# Obtain the current fan speed
try:
fan_speed = sysdata.fanSpeed
except Exception:
fan_speed = None
# Obtain the NVMe device temperature
try:
nvme_c = sysdata.driveTemp
except Exception:
nvme_c = None
# Obtain the NVMe Device read and write rates
try:
read_mb, write_mb = sysdata.driveStats
read_mb = float(read_mb)
@@ -202,13 +352,12 @@ class MonitorWindow(QMainWindow):
except Exception:
read_mb, write_mb = None, None
# Get the CPU load precentages
try:
p = cpuload.getPercentages()
values = []
for i in range( len(cpuload) ):
values.append( round( p[f'cpu{i}'], 2 ) )
values = [p[name] for name in cpuload.cpuNames]
except Exception:
values = [ None for i in range( len( cpuload) ) ]
values = [ None for name in cpuload.cpuNames ]
# Append to charts
self.cpu_chart.append([cpu_c,nvme_c])

View File

@@ -113,7 +113,6 @@ class systemData:
def CPUTemperature(self) -> int:
return self._cpuTemp.temperature
@property
def fanSpeed( self ) -> int:
speed= 0
@@ -157,8 +156,8 @@ class systemData:
@property
def driveStats(self) -> tuple[float,float]:
data = self._stats.readWriteSectors()
readMB = (float(data[0]) * 512.0) / (1024.0 * 1024.0)
writeMB = (float(data[1]) * 512.0) / (1024.0 * 1024.0)
readMB = (float(data[0]) * 512.0) #/ (1024.0 * 1024.0)
writeMB = (float(data[1]) * 512.0) #/ (1024.0 * 1024.0)
return (readMB, writeMB )

273
monitor/systemsupport.py Executable file
View File

@@ -0,0 +1,273 @@
#!/usr/bin/python3
#
# Setup environment and pull in all of the items we need from gpiozero. The
# gpiozero library is the new one that supports Raspberry PI 5's (and I suspect
# will be new direction for all prior version the RPIi.)
#
from gpiozero import CPUTemperature
import time
import os
class DriveStats:
'''
DriveStat class -
This class gets the drive statistics from sysfs for the device passed
in. There are several statistics can can be obtained. Note that since
all of the data is pulled at the same time, it is upto the caller to
make sure all the stats needed are obtained at the same time.
See: https://www.kernel.org/doc/html/latest/block/stat.html
Parameters:
device - the name of the device to track
'''
READ_IOS = 0
READ_MERGES = 1
READ_SECTORS = 2
READ_TICKS = 3
WRITE_IOS = 4
WRITE_MERGES = 5
WRITE_SECTORS = 6
WRITE_TICKS = 7
IN_FLIGHT = 8
IO_TICKS = 9
TIME_IN_QUEUE = 10
DISCARD_IOS = 11
DISCARD_MERGES = 12
DISCARD_SECTORS = 13
DISCARD_TICS = 14
FLUSH_IOS = 15
FLUSH_TICKS = 16
def __init__( self, device:str ):
self._last : list[int] = []
self._stats : list[int] = []
self._device = device
self._readStats()
def _readStats( self ):
'''
Read the disk statistics. The stored statics in sysfs are stored as a single file
so that when the data is read, all of the stats correlate to the same time. The data
is from the time the device has come online.
last and set to the old version of the data, and the latest data is stored in stats
'''
try:
self._last = self._stats
with open( f"/sys/block/{self._device}/stat", "r") as f:
curStats = f.readline().strip().split(" ")
self._stats = [int(l) for l in curStats if l]
except Exception as e:
print( f"Failure reading disk statistics for {self._device} error {e}" )
def _getStats( self ) -> list[int]:
'''
Read the devices statistics from the device,and return it.
Returns:
An array containing all of the data colleected about the device.
'''
curData : list[int] = []
self._readStats()
if self._last == []:
curData = self._stats[:]
else:
curData = [ d-self._last[i] for i,d in enumerate( self._stats ) ]
return curData
def readAllStats( self ) -> list[int]:
'''
read all of the drive statisics from sysfs for the device.
Returns
A list of all of the device stats
'''
return self._getStats()
def readSectors( self )-> int:
return self._getStats()[DriveStats.READ_SECTORS]
def writeSectors( self ) -> int:
return self._getStats()[DriveStats.WRITE_SECTORS]
def discardSectors( self ) -> int:
return self._getStats()[DriveStats.DISCARD_SECTORS]
def readWriteSectors( self ) -> tuple[int,int]:
curData = self._getStats()
return (curData[DriveStats.READ_SECTORS],curData[DriveStats.WRITE_SECTORS])
class systemData:
def __init__( self, drive : str = 'nvme0n1' ):
self._drive = drive
self._cpuTemp = CPUTemperature()
self._stats = DriveStats( self._drive )
@property
def CPUTemperature(self) -> int:
return self._cpuTemp.temperature
@property
def fanSpeed( self ) -> int:
speed= 0
try:
command = os.popen( 'cat /sys/devices/platform/cooling_fan/hwmon/*/fan1_input' )
speed = int( command.read().strip())
except Exception as error:
print( f"Could not determine fan speed, error {error}" )
finally:
command.close()
return speed
@property
def driveTemp(self) -> float:
smartOutRaw = ""
cmd = f'sudo smartctl -A /dev/{self._drive}'
try:
command = os.popen( cmd )
smartOutRaw = command.read()
except Exception as error:
print( f"Could not launch {cmd} error is {error}" )
return 0.0
finally:
command.close()
smartOut = [ l for l in smartOutRaw.split('\n') if l]
for smartAttr in ["Temperature:","194","190"]:
try:
line = [l for l in smartOut if l.startswith(smartAttr)][0]
parts = [p for p in line.replace('\t',' ').split(' ') if p]
if smartAttr == "Temperature:":
return float(parts[1])
else:
return float(parts[0])
except IndexError:
pass
return float(0.0)
@property
def driveStats(self) -> tuple[float,float]:
data = self._stats.readWriteSectors()
readMB = (float(data[0]) * 512.0) #/ (1024.0 * 1024.0)
writeMB = (float(data[1]) * 512.0) #/ (1024.0 * 1024.0)
return (readMB, writeMB )
class CPULoad:
'''
A class to help with obtaining the CPU load of the system. If there is more information
needed, we can add to this.
Note:
This code automatically attempts to load the data from the system to initialize the
object with names, and an initial set of data.
This may result in th first actual call return some not very consistent values, for
the time period being observed, but that difference is minimal. In otherwords if we
the period of time being measured is 1 second, and it's been a minute since this class
was initialized, the first period reported will be CPU load over the minute, not 1 second,
and the second period reported will be for a second...
This is usually not an issue.
'''
def __init__( self ):
self._previousData : dict[str,tuple] = self._getRawData()
self._names : list[str] = []
for item in self._previousData:
self._names.append( item )
def _getRawData( self ) -> dict[str : tuple]:
'''
Obtain the raw CPU data from the system (located in /prop/stat), and
return just the cpu0 -> cpux values. No assumption is made on the number of
cpus.
Returns:
A dictionary is returned, the format is name = (total, idle). The total
time and idle time are use to determine the percent utilization of the system.
'''
result = {}
with open( "/proc/stat", "r") as f:
allLines = f.readlines()
for line in allLines:
cpu = line.replace('\t', ' ').strip().split()
if (len(cpu[0]) > 3) and (cpu[0][:3] == "cpu"):
total = 0
idle = 0
for i in range( 1, len(cpu)):
total += int(cpu[i])
if i == 4 or i == 5:
idle += int(cpu[i])
result[cpu[0]] = (total,idle)
return result
def getPercentages( self ) -> dict[ str : float ]:
'''
Obtain the percent CPU utilization of the system for a period of time.
This routine gets the current raw data from the system, and then performs
a delta from the prior time this function was called. This data is then run
through the following equation:
utilization = ((total - idle)/total) * 100
If the snapshots are taken at relativy consistent intervals, the CPU
utilization in percent, is reasonably lose to the actual percentage.
Returns:
A dictionary consisting of the name of the CPU, and a floating point
number representing the current utilization of that CPU.
'''
results = {}
current = self._getRawData()
for item in current:
total = current[item][0] - self._previousData[item][0]
idle = current[item][1] - self._previousData[item][1]
percent = ((total - idle)/total) * 100
results[item] = round(percent,2)
self._previousData = current
return results
@property
def cpuNames( self ) -> list[str]:
'''
Get a list of CPU names from the system.
Returns:
a list of strings
'''
return self._names
def __len__(self) -> int:
'''
handle getting the length (or count of CPU's).
Returns:
Number of CPU's
'''
return len(self._previousData)
if __name__ == "__main__":
data = systemData()
print( f"CPU Temp : {data.CPUTemperature}" )
print( f"Fan Speed: {data.fanSpeed}" )
print( f"NVME Temp: {data.driveTemp}" )
print( f"Stats : {data.driveStats}" )
load = CPULoad()
print( f"Number of CPU's = {len(load)}" )
for i in range(10):
time.sleep( 1 )
percentage : dict[str:float] = load.getPercentages()
print( f"percentage: {percentage}" )
for item in percentage:
print( f"{item} : {percentage[item]:.02f}" )