def _collect_inodes_metrics(self, mountpoint):
metrics = {}
# we need to timeout this, too.
try:
inodes = timeout(5)(os.statvfs)(mountpoint)
except TimeoutException:
self.log.warning(
u'Timeout while retrieving the disk usage of `%s` mountpoint. Skipping...',
mountpoint)
return metrics
except Exception as e:
self.log.warning('Unable to get disk metrics for %s: %s',
mountpoint, e)
return metrics
if inodes.f_files != 0:
total = inodes.f_files
free = inodes.f_ffree
metrics[self.METRIC_INODE.format('total')] = total
metrics[self.METRIC_INODE.format('free')] = free
metrics[self.METRIC_INODE.format('used')] = total - free
# FIXME: 6.x, use percent, a lot more logical than in_use
metrics[self.METRIC_INODE.format('in_use')] = (total -
free) / total
return metrics
def _collect_inodes_metrics(self, mountpoint):
metrics = {}
# we need to timeout this, too.
try:
inodes = timeout(self._timeout)(os.statvfs)(mountpoint)
except TimeoutException:
self.log.warning(
u'Timeout after %d seconds while retrieving the disk usage of `%s` mountpoint. '
u'You might want to change the timeout length in the settings.',
self._timeout,
mountpoint,
)
return metrics
except Exception as e:
self.log.warning(
u'Unable to get disk metrics for %s: %s. '
u'You can exclude this mountpoint in the settings if it is invalid.',
mountpoint,
e,
)
return metrics
if inodes.f_files != 0:
total = inodes.f_files
free = inodes.f_ffree
metrics[self.METRIC_INODE.format('total')] = total
metrics[self.METRIC_INODE.format('free')] = free
metrics[self.METRIC_INODE.format('used')] = total - free
# FIXME: 8.x, use percent, a lot more logical than in_use
metrics[self.METRIC_INODE.format('in_use')] = (total -
free) / total
return metrics
def collect_metrics_psutil(self):
self._valid_disks = {}
for part in psutil.disk_partitions(all=True):
# we check all exclude conditions
if self.exclude_disk(part):
continue
# Get disk metrics here to be able to exclude on total usage
try:
disk_usage = timeout(5)(psutil.disk_usage)(part.mountpoint)
except TimeoutException:
self.log.warning(
u'Timeout while retrieving the disk usage of `%s` mountpoint. Skipping...', part.mountpoint
)
continue
except Exception as e:
self.log.warning('Unable to get disk metrics for %s: %s', part.mountpoint, e)
continue
# Exclude disks with total disk size 0
if disk_usage.total == 0:
continue
# For later, latency metrics
self._valid_disks[part.device] = (part.fstype, part.mountpoint)
self.log.debug('Passed: {}'.format(part.device))
device_name = part.mountpoint if self._use_mount else part.device
tags = [part.fstype, 'filesystem:{}'.format(part.fstype)] if self._tag_by_filesystem else []
tags.extend(self._custom_tags)
# apply device/mountpoint specific tags
for regex, device_tags in self._device_tag_re:
if regex.match(device_name):
tags.extend(device_tags)
if self.devices_label.get(device_name):
tags.append(self.devices_label.get(device_name))
# legacy check names c: vs psutil name C:\\
if Platform.is_win32():
device_name = device_name.strip('\\').lower()
tags.append('device:{}'.format(device_name))
for metric_name, metric_value in iteritems(self._collect_part_metrics(part, disk_usage)):
self.gauge(metric_name, metric_value, tags=tags)
# Add in a disk read write or read only check
if self._service_check_rw:
rwro = {'rw', 'ro'} & set(part.opts.split(','))
if len(rwro) == 1:
self.service_check(
'disk.read_write', AgentCheck.OK if rwro.pop() == 'rw' else AgentCheck.CRITICAL, tags=tags
)
else:
self.service_check('disk.read_write', AgentCheck.UNKNOWN, tags=tags)
self.collect_latency_metrics()
def check(self, instance):
"""Get disk space/inode stats"""
if self._tag_by_label and Platform.is_linux():
self.devices_label = self._get_devices_label()
self._valid_disks = {}
for part in psutil.disk_partitions(all=self._include_all_devices):
# we check all exclude conditions
if self.exclude_disk(part):
continue
# Get disk metrics here to be able to exclude on total usage
try:
disk_usage = timeout(self._timeout)(psutil.disk_usage)(
part.mountpoint)
except TimeoutException:
self.log.warning(
u'Timeout after %d seconds while retrieving the disk usage of `%s` mountpoint. '
u'You might want to change the timeout length in the settings.',
self._timeout,
part.mountpoint,
)
continue
except Exception as e:
self.log.warning(
u'Unable to get disk metrics for %s: %s. '
u'You can exclude this mountpoint in the settings if it is invalid.',
part.mountpoint,
e,
)
continue
# Exclude disks with size less than min_disk_size
if disk_usage.total <= self._min_disk_size:
if disk_usage.total > 0:
self.log.info(
'Excluding device %s with total disk size %s',
part.device, disk_usage.total)
continue
# For later, latency metrics
self._valid_disks[part.device] = (part.fstype, part.mountpoint)
self.log.debug('Passed: %s', part.device)
device_name = part.mountpoint if self._use_mount else part.device
tags = [part.fstype, 'filesystem:{}'.format(part.fstype)
] if self._tag_by_filesystem else []
tags.extend(self._custom_tags)
# apply device/mountpoint specific tags
for regex, device_tags in self._device_tag_re:
if regex.match(device_name):
tags.extend(device_tags)
if self.devices_label.get(device_name):
tags.extend(self.devices_label.get(device_name))
# legacy check names c: vs psutil name C:\\
if Platform.is_win32():
device_name = device_name.strip('\\').lower()
tags.append('device:{}'.format(device_name))
tags.append('device_name:{}'.format(_base_device_name(
part.device)))
for metric_name, metric_value in iteritems(
self._collect_part_metrics(part, disk_usage)):
self.gauge(metric_name, metric_value, tags=tags)
# Add in a disk read write or read only check
if self._service_check_rw:
rwro = {'rw', 'ro'} & set(part.opts.split(','))
if len(rwro) == 1:
self.service_check('disk.read_write',
AgentCheck.OK if rwro.pop() == 'rw' else
AgentCheck.CRITICAL,
tags=tags)
else:
self.service_check('disk.read_write',
AgentCheck.UNKNOWN,
tags=tags)
self.collect_latency_metrics()
