0768a400a1
Fix: Issue 157: Document how to troubleshoot data inconsistencies |
||
---|---|---|
.circleci | ||
.github | ||
systemd | ||
testdata | ||
.gitignore | ||
.golangci.yml | ||
.promu.yml | ||
.yamllint | ||
CHANGELOG.md | ||
CODE_OF_CONDUCT.md | ||
collect-smartctl-json.sh | ||
device_filter.go | ||
device_filter_test.go | ||
Dockerfile | ||
EXAMPLE.md | ||
gjsonext.go | ||
go.mod | ||
go.sum | ||
LICENSE | ||
main.go | ||
MAINTAINERS.md | ||
Makefile | ||
Makefile.common | ||
metrics.go | ||
readjson.go | ||
README.md | ||
redact_fake_json.py | ||
SECURITY.md | ||
smartctl.go | ||
smartctlinfo.go | ||
VERSION |
smartctl_exporter
Export smartctl statistics to prometheus
Example output you can show in EXAMPLE.md
Need more?
If you need additional metrics - contact me :)
Create a feature request, describe the metric that you would like to have and attach exported from smartctl json file
Requirements
smartmontools
>= 7.0, because export to json released in 7.0
Configuration
Command line options
The exporter will scan the system for available devices if no --smartctl.device
flags are used.
usage: smartctl_exporter [<flags>]
Flags:
-h, --help Show context-sensitive help (also try --help-long and --help-man).
--smartctl.path="/usr/sbin/smartctl"
The path to the smartctl binary
--smartctl.interval=60s The interval between smartctl polls
--smartctl.rescan=10m The interval between rescanning for new/disappeared devices. If the interval is smaller than 1s no
rescanning takes place. If any devices are configured with smartctl.device also no rescanning takes
place.
--smartctl.device=SMARTCTL.DEVICE ...
The device to monitor (repeatable)
--smartctl.device-exclude=""
Regexp of devices to exclude from automatic scanning. (mutually exclusive to
device-include)
--smartctl.device-include=""
Regexp of devices to include in automatic scanning. (mutually exclusive to
device-exclude)
--web.telemetry-path="/metrics"
Path under which to expose metrics
--web.systemd-socket Use systemd socket activation listeners instead of port listeners (Linux only).
--web.listen-address=:9633 ...
Addresses on which to expose metrics and web interface. Repeatable for multiple
addresses.
--web.config.file="" [EXPERIMENTAL] Path to configuration file that can enable TLS or authentication.
--log.level=info Only log messages with the given severity or above. One of: [debug, info, warn,
error]
--log.format=logfmt Output format of log messages. One of: [logfmt, json]
--version Show application version.
TLS and basic authentication
This exporter supports TLS and basic authentication.
To use TLS and/or basic authentication, you need to pass a configuration file
using the --web.config.file
parameter. The format of the file is described
in the exporter-toolkit repository.
Example of running in Docker
Minimal functional docker-compose.yml
:
version: "3"
services:
smartctl-exporter:
image: prometheuscommunity/smartctl-exporter
privileged: true
user: root
ports:
- "9633:9633"
Troubleshooting
Troubleshooting data inconsistencies
smartmon_exporter
uses the JSON output from smartctl
to provide the data to
Prometheus. If the data is incorrect, look at the data from smartctl
to
determine if the issue should be reported upstream to smartmontools or to this
repo. In general, the smartctl_exporter
should not modify the data in flight.
If the data is missing from smartctl
, it should not be in smartctl_exporter
.
If the data from smartctl
is incorrect, it should be reported upstream.
Requests for smartctl_exporter
to "fix" incorrect data where smartctl
is
reporting incorrect data will be closed. The grey area is when invalid or
missing data from smartctl is causing multiple invalid or incorrect data
in smartctl_exporter
. This could happen if the data is used in a calculation
for other data. This will need to be researched on a case by case basis.
- | smartctl valid | smartctl missing | smartctl invalid/incorrect |
---|---|---|---|
smartctl_exporter valid | all good | N/A | N/A |
smartctl_exporter missing | issue for smartctl_exporter | report upstream to smartmontools | report upstream to smartmontools |
smartctl_exporter invalid | issue for smartctl_exporter | issue for smartctl_exporter and report upstream | report upstream to smartmontools |
smartctl output vs smartctl_exporter output
The S.M.A.R.T. attributes are mapped in
smartctl.go.
Each function has a prometheus.MustNewConstMetric
or similar function with the
first parameter being the metric name. Find the metric name in
metrics.go
to see how the exporter displays the information. This may sound technical, but
it's crucial for understanding how data flows from smartctl
to
smartctl_exporter
to Prometheus.
If the data looks incorrect, check the Smartmontools Frequently Asked Questions (FAQ). It's likely your question may already have an answer. If you still have questions, open an issue.
Gathering smartctl data
Follow these steps to gather smartctl data for troubleshooting purposes. If you have unique drives/data/edge cases and would like to "donate" the data, open a PR with the redacted JSON files.
- Run
scripts/collect-smartctl-json.sh
to export all drives to asmartctl-data
directory (created in the current directory). - Run
scripts/redact_fake_json.py
to redact sensitive data. - Provide the JSON file for the drive in question.
cd scripts
./collect-smartctl-json.sh
./redact-fake-json.py smartctl-data/*.json
Run smartctl_exporter using JSON data
The smartctl_exporter
can be run using local JSON data. The device names are
pulled from actual devices in the machine while the data is redirected to the
debug
directory. Save the JSON data in the debug
directory using the actual
device names using a 1:1 ratio. If you have 3 devices, sda
, sdb
and sdc
,
the smartctl_exporter
will expect 3 files: debug/sda.json
, debug/sdb.json
and debug/sdc.json
.
Once the "fake devices" (JSON files) are in place, run the exporter passing the
hidden --smartctl.fake-data
switch on the command line. The port is specified
to prevent conflicts with an existing smartctl_exporter
on the default port.
smartctl_exporter --web.listen-address 127.0.0.1:19633 --smartctl.fake-data
FAQ
How do I run smartctl_exporter
against a JSON file?
If you're helping someone else, request the output of the smartctl
command
above. Feed this into the smartctl_exporter
using the
hidden --smartctl.fake-data
flag. If a smartctl_exporter
is already running,
use a different port; in this case, it's 19633
. Run collect_fake_json.sh
first to collect the JSON files for your devices. Copy the requested JSON
file into one of the fake files. After starting the exporter, you can query it
to see the data generated.
# Dump the JSON files for your devices into debug/
./collect_fake_json.sh
# copy the test JSON into one of the files in debug/
cp extracted-from-above-sda.json debug/sda.json
# Make sure you have the latest version
go build
# Use a different port in case smartctl_exporter is already running
sudo ./smartctl_exporter --web.listen-address=127.0.0.1:19633 --log.level=debug --smartctl.fake-data
# Use curl with grep
curl --silent 127.0.0.1:19633/metrics | grep -i nvme
# Or xh with ripgrep
xh --body :19633/metrics | rg nvme
Why is root required? Can't I add a user to the "disk" group?
A blogger had the same question and opened a ticket on smartmontools. This is
their response. smartctl
needs to be run as root.
RFE: add O_RDRW mode for sat/scsi/ata devices
According to function
blk_verify_command()
from current kernel sources (see block/scsi_ioctl.c), O_RDONLY or O_RDWR make no difference if device was opened as root (or with CAP_SYS_RAWIO).The SCSI commands listed in function
blk_set_cmd_filter_defaults()
show that some of thesmartctl -d scsi
functionality might work with O_RDONLY for non-root users. Some more might work with O_RDWR.But
smartctl -d sat
(to access SATA devices) won't work at all because the SCSI commands ATA_12 and ATA_16 (see scsi_proto.h) are always blocked for non-root users.
What about my NVMe drive?
From the smartmontools FAQ: My NVMe drive is not in the smartctl/smartd database
SCSI/SAS and NVMe drives do not provide ATA/SATA-like SMART Attributes. Therefore the drive database does not contain any entries for these drives. This may change in the future as some drives provide similar info via vendor specific commands (see ticket #870).
smartmontools also has a wiki page for NVMe devices.
How do I report upstream to smartmontools?
Check their FAQ: How to create a bug report.