This is a monitoring/inventory plugin to check components and health status of systems which support Redfish. It will also create an inventory of all components of a system.
You are also able to import the inventory files into NetBox using netbox-snyc.
- python >= 3.6
- redfish >= 2.1.4
- on RedHat/CentOS you need to install python3.6 and pip from EPEL first
- on RedHat/CentOS 8 systems the package name changed to
python3-pip
yum install python36-pip
- download and install plugin
cd /usr/lib64/nagios/plugins/
git clone https://github.com/bb-Ricardo/check_redfish.git
cd check_redfish
pip3 install -r requirements.txt || pip install -r requirements.txt
Command definitions and a service config example for Icinga2 can be found in contrib. There is also an InfluxDB dashboard for some metrics included.
usage: check_redfish.py [-H HOST] [-u USERNAME] [-p PASSWORD] [-f AUTHFILE]
[--sessionfile SESSIONFILE]
[--sessionfiledir SESSIONFILEDIR] [--nosession] [-h]
[-w WARNING] [-c CRITICAL] [-v] [-d] [-m MAX]
[-r RETRIES] [-t TIMEOUT] [--storage] [--proc]
[--memory] [--power] [--temp] [--fan] [--nic] [--bmc]
[--info] [--firmware] [--sel] [--mel] [--all] [-i]
[--inventory_id INVENTORY_ID]
[--inventory_file INVENTORY_FILE]
This is a monitoring/inventory plugin to check components and
health status of systems which support Redfish.
It will also create a inventory of all components of a system.
R.I.P. IPMI
Version: 1.3.2 (2022-01-23)
mandatory arguments:
-H HOST, --host HOST define the host to request. To change the port just
add ':portnumber' to this parameter
authentication arguments:
-u USERNAME, --username USERNAME
the login user name
-p PASSWORD, --password PASSWORD
the login password
-f AUTHFILE, --authfile AUTHFILE
authentication file with user name and password
--sessionfile SESSIONFILE
define name of session file
--sessionfiledir SESSIONFILEDIR
define directory where the plugin saves session files
--nosession Don't establish a persistent session and log out after
check is finished
optional arguments:
-h, --help show this help message and exit
-w WARNING, --warning WARNING
set warning value
-c CRITICAL, --critical CRITICAL
set critical value
-v, --verbose this will add all https requests and responses to
output, also adds inventory source data to all
inventory objects
-d, --detailed always print detailed result
-m MAX, --max MAX set maximum of returned items for --sel or --mel
-r RETRIES, --retries RETRIES
set number of maximum retries (default: 3)
-t TIMEOUT, --timeout TIMEOUT
set number of request timeout per try/retry (default:
7)
query status/health information (at least one is required):
--storage request storage health
--proc request processor health
--memory request memory health
--power request power supply health
--temp request temperature sensors status
--fan request fan status
--nic request network interface status
--bmc request bmc info and status
--info request system information
--firmware request firmware information
--sel request System Log status
--mel request Management Processor Log status
--all request all of the above information at once
query inventory information (no health check):
-i, --inventory return inventory in json format instead of regular
plugin output
--inventory_id INVENTORY_ID
set an ID which can be used to identify this host in
the destination inventory
--inventory_file INVENTORY_FILE
set file to write the inventory output to. Otherwise
stdout will be used.
multiple request commands can be combined. Or use --all to query all system information at once
/usr/lib64/nagios/plugins/check_redfish/check_redfish.py -H 10.0.0.23 -f /etc/icinga2/ilo_credentials --storage --power
- request BMC: 10.0.0.23
- use credentials from file: /etc/icinga2/ilo_credentials
- query health of: storage, power
If you want to use a different Port then 443 then just add the port to the Host parameter.
Example for Port 8443:
-H 127.0.0.1:8443
Credentials can be provided in 3 ways and will be checked in following order:
- using --username and --password options (credentials will be exposed in process list)
- using an auth file which includes the credentials
- add credentials to environment
An authentication credential file can be provided. The structure looks like this:
username=icinga
password=readonlysecret
these two environment vars will be checked
- CHECK_REDFISH_USERNAME
- CHECK_REDFISH_PASSWORD
export CHECK_REDFISH_USERNAME=icinga
export CHECK_REDFISH_PASSWORD=readonlysecret
To avoid delays due to login on every request and flooding the event log with login/logout messages a session resumption was implemented. If the session in the BMC is expired a new session and session file will be created.
IMPORTANT
To actually benefit from this feature you need to set the user session timeout in
the BMC to a higher value then your default check interval!
If your default check interval is 5 minutes then the session timeout in the BMC should be at least 6 minutes!
If no session is required (i.e.: testing, inventory collection) then a --nosession can
be added to close session on the BMC properly.
Per default a session file will be crated in the system/user default temp path. These defaults can be changed with following options:
Use --sessionfiledirto define where the session files should be stored.
Use --sessionfile to specify the name of the session file for this particular system.
options like this:
--sessionfiledir /var/plugin/tmp --sessionfile my-hostname.session
results in following session file:
/var/plugin/tmp/my-hostname.session
you can use warning and critical with following commands:
--mel and --sel (values are passed as "days")
define after how many days' event log entries which have a != OK severity shouldn't
be alerted anymore. On most systems it is not possible to set management event log entries
as cleared. So entries with a severity of warning would alarm forever. This way they change
state while they age.
Example: --mel --critical 1 --warning 3
- Entries with a != OK severity which are not older than 24 hours are reported as CRITICAL
- Entries with a != OK severity which are not older than 72 hours are reported as WARNING
- Any entries with a != OK severity which are older than 72 hours will be reported as OK
Health status by default will be reported as a summary:
[OK]: All power supplies (2) are in good condition|'ps_1'=122 'ps_2'=109
If multiline output by default is preferred the option --detailed needs to be added
[OK]: Power supply 1 (865408-B21) status is: Ok
[OK]: Power supply 2 (865408-B21) status is: Ok|'ps_1'=121 'ps_2'=109
Use option --verbose to check for connection problems.
All redfish https requests and responses will be printed.
This option can be used to limit the results output for event log entries requested by --mel and --sel
Sometimes an iLO4 BMC can be very slow in answering Redfish request. To avoid getting "retries exhausted"
alarms you can increase the number of retries and/or the timeout. The timeout defines the seconds after each
try/retry times out. If you increase these values make sure to also adjust the check_timeout setting
in your Icinga2 service definition. The total runtime of
this plugin (if all retries fail) can be calculated like this: (1. try + num retries) * timeout
The default number of retries is set to 3 and the default timeout is set to 7. In case all retries fail then the plugin would be finished after 28 seconds.
(1 + 3) * 7 = 28
This plugin is able to return a (almost) complete inventory of the queried system.
Just add the command option --inventory or -i to get the inventory
in a JSON format.
IMPORTANT
This is the first official version and might still change later on. If you encounter problems or have
suggestions for changes/improvements then please create a GitHub issue.
{
"inventory": {
"chassi": [],
"fan": [],
"firmware": [],
"logical_drive": [],
"manager": [],
"memory": [],
"network_adapter": [],
"network_port": [],
"physical_drive": [],
"power_supply": [
{
"bay": 1,
"capacity_in_watt": 500,
"chassi_ids": [
1
],
"firmware": "1.03",
"health_status": "OK",
"id": "0",
"input_voltage": 224,
"last_power_output": 110,
"model": "XXXXXX-B21",
"name": "HpeServerPowerSupply",
"operation_status": "Enabled",
"part_number": "XXXXXX-001",
"serial": "XXXXXXX",
"type": "AC",
"vendor": "CHCNY"
},
{
"bay": 2,
"capacity_in_watt": 500,
"chassi_ids": [
1
],
"firmware": "1.03",
"health_status": "OK",
"id": "1",
"input_voltage": 228,
"last_power_output": 110,
"model": "XXXXXX-B21",
"name": "HpeServerPowerSupply",
"operation_status": "Enabled",
"part_number": "XXXXXX-001",
"serial": "XXXXXXX",
"type": "AC",
"vendor": "CHCNY"
}
],
"processor": [],
"storage_controller": [],
"storage_enclosure": [],
"system": [],
"temperature": []
},
"meta": {
"data_retrieval_issues": {},
"duration_of_data_collection_in_seconds": 1.002901,
"host_that_collected_inventory": "inventory-collector.example.com",
"inventory_id": null,
"inventory_layout_version": "1.2.0",
"script_version": "1.2.0",
"start_of_data_collection": "2021-04-01T09:09:07+02:00"
}
}
In case you need more information or want to debug the data you can add the verbose
option. This will also add the source_data attribute for each inventory item.
You can find a list of attributes for each item here
It is also possible to use the cli option --inventory_file to write the inventory data to a file.
This way it can be forwarded or used in an inventory import tool. Here you also might want to use
--inventory_id to get a fixed reference to an existing object.
- On HPE iLO4 a maximum of 30 entries will be returned for the commands --mel and --sel
- On almost all systems the nic status is reported unreliable
- For --storage components which report a Status.Health as None will be treated as OK if Status.State is set to Enabled
This plugin is currently tested with following systems
Almost all Server which have iLO4 (>=2.50) or iLO5 (>=1.20) should work
- ProLiant BL460c Gen8
- ProLiant BL460c Gen9
- ProLiant BL460c Gen10
- ProLiant DL360p Gen8
- ProLiant DL360 Gen9
- ProLiant DL360 Gen10
- ProLiant DL380p Gen8
- ProLiant DL380 Gen9
- ProLiant DL380 Gen10
- ProLiant DL560 Gen10
- ProLiant DL580 Gen8
- ProLiant DL580 Gen9
- ThinkSystem SR650 (BMC Version 2.12)
- PowerEdge R630 (iDRAC 8 Version 2.70.70.70)
- PowerEdge R640 (iDrac 9 Version 4.22.00.00)
- PowerEdge R740 (iDRAC 9 Version 3.32.32.32)
- PowerEdge R740xd (iDRAC 9 Version 4.00.00.00)
- PowerEdge R7515 (iDRAC 9 Version 4.10.10.10)
- PowerEdge R840 (iDRAC 9 Version 4.22.00.00)
- PowerEdge R930 (iDRAC 8 Version 2.70.70.70)
- TaiShan 2280 V2 (iBMC Version 3.63)
- X8600 Blade (iBMC Version 3.04)
IMPORTANT:
- iRMC S5 Firmware 2.60 till 2.63 has a minor bug, avoid usage!
- make sure to use
--detailedoption for memory check. HealthRollup for Memory is not working/existent! - iRMC S5 Firmware 3.06P reports storage data twice for some components Models:
- PRIMERGY RX2530 M5 (iRMC Version 2.50P)
- PRIMERGY RX2540 M4 (iRMC Version 2.50P)
- PRIMERGY RX2540 M5 (iRMC Version 2.50P)
- Cisco C220M4SX (CIMC Version 4.1(2a))
- Cisco C220M5SX (CIMC Version 3.1(3a))
- Cisco C240M5SX (CIMC Version 3.1(3a))
- Inspur NF5280 M5 (4.26.3)
- Inspur NF5280 M6 (4.12.04)
You can check out the full license here
This project is licensed under the terms of the MIT license.