Software interface

There are several software interfaces available to monitor the status of the RECS®|Box system. These are the Management WebGUI, a REST API providing XML based monitoring and management functionality and a native NRPE based Nagios interface.

Management WebGUI

The Management WebGUI is established on every RECS®|Box unit. Accessible by any known browser on the assigned IP address and the default port 80. The following views are dependent on the device and assembly.

In general these symbols have the following meaning on every page:

Everything is OK. Also indicated by a green line in a graph.
Warnung. Something is wrong, but the system is still fully functional. The system has to be checked so the problem doesn't get worse. Indicated by a yellow line in a graph.
Critical Error. The system must be checked immediately and maybe has to be shut down to prevent hardware damage. indicated by a red line in a graph.

Figure 1 shows the first call of the Management WebGUI. It is organized into three columns. The first is on the left-hand side and contains the following:

Overview: General overview of all managed RCUs, RPUs, installed nodes and health status
Management: Selection of every managed RCU and RPU in the rack with a sensor view button for the Arneb
Global settings: IP filter and firmware update
Log: Logs from the management software about system health and java messages. The logs can be downloaded as a zipfile

The second colum contains the buttons and sliders to manipulate the system. While the third colum is mostly for history information like power usage and temperature graphs.

Overview

All units that are installed in the rack and that are managed by the software are summarized on this page. The total power usage is summed up over all managed units.

Management OverviewFig. 1

Management

An overview of the selected unit can be seen in this tab. The fans can be regulated by dragging the slider to the desired percentage. And multiple nodes can be selected. By klicking on a node theNode management page of the node is shown.

Adds an ImageCaption tagFig. 2

A quick menu to control a node can be opened by klicking on the gear next to an CXP node. In this menu the node can be switched on and off and the KVM can be switched to the node.

Management pop-pu for Apalis nodesFig. 3

Apalis nodes do not show a management pop-up button due to size constraints.
Click on the node button while pressing the “Shift” key to open the management pop-up instead of navigating to the node view.

When pressing the “Shift” key while clicking, the “Select all” and “Select none” buttons select only nodes currently on or nodes currently off, respectively.

Node management

On this page the selected node can be controlled and detailed status values and graphs can be seen.
By klicking on the arrow, pointing downwards in the upper bar next to the nodename, the other nodes of the unit can be chosen.

Node managementFig. 4

Global settings

All IPs that are allowed to access the Nagios interface have to be listet here.
The firmware for the whole RECS®|Box can be uploaded here by klicking on the “Upload Firmware File” button and selecting the file. The update-process starts right after the file was uploaded.
For the update process all modules will be powered off!

Golobal settings tabFig. 5

Log viewer

In the system healths tab of the log page the status changes of the sensors, fan and boards can be seen.

System health logFig. 6

In the java tab of the log page all messages regarding the software can be found.

Java messagesFig. 7

Several filters can be set for both tabs at the top.
At the bottom the whole log can be downloaded as a ZIP file containing the individual logfiles.

Redfish API

The documentation of the RECS®|Box Redfish API can be seen at Github.

REST API

Access

The RECS®|Box Management API is accessible via the IP-Address or the hostname of the TOR-Master of the cluster. The basic URL of the API has the format https://TOR-Master/REST/ or http://TOR-Master/REST/.

Accessing the REST API requires HTTP Basic authentication. The authenticated user has to be in the “Admin” or “User” group to be able to execute the POST/PUT management calls.

Components

The RECS®|Box Management API makes all hardware components in the cluster available as XML trees in software. The following components are supported by the API:

Attribute Description
node A single node
backplane A backplane can be equipped with zero or more baseboards
baseboard A baseboard can be equipped with zero or more nodes
rcu A RECS®|Box Computing Unit (RCU) can be equipped with zero or more baseboards
rack A rack consists of several RCUs

Many resources also return lists of components. These are named according to the scheme <component name>List (e.g. nodeList, rcuList) and contain the elements of the list.

Example of a backplaneList:

<backplaneList>
<backplane rcuPosition="1" id="RCU_84055620466592_BP_1" infrastructurePower="0.0">
<temperatures>24.0</temperatures>
<temperatures>25.0</temperatures>
<temperatures>26.0</temperatures>
<temperatures>27.0</temperatures>
<temperatures>28.0</temperatures>
</backplane>
</backplaneList>

Node

Example XML:

<node baseboardPosition="0" maxPowerUsage="44" actualNodePowerUsage="32.426884399865166" 
actualPEGPowerUsage="15.12053962324833" actualPowerUsage="47.54742402311349" architecture="x86" 
baseboardId="RCU_84055620466592_BB_1" health="OK" id="RCU_84055620466592_BB_1_0" inletTemperature="20.0" 
lastSensorUpdate="1465470151268" macAddressCompute="70:b3:d5:56:40:48" outletTemperature="20.0" state="1" 
highestTemperature="20.0" voltage="12.072700851453936"/>

The following table shows the possible attributes (some are optional) and their meaning:

Attribute Description Unit Data type
idUnique ID for referencing the component-String
actualPowerUsage Actual power consumption of a node (Node + PEG)WDouble
actualNodePowerUsage Actual power consumption of a node (Node only)WDouble
actualPEGPowerUsage Actual power consumption of a PEG cardWDouble
maxPowerUsage Maximum power the node can drawWInteger
baseboardId ID of the baseboard which hosts the node-String
baseboardPosition Position of the node on the baseboard-Integer
state Power state of the node (0=Off, 1=On, 2=Soft-off, 3=Standby, 4=Hibernate)-Integer
architecture Architecture (x86, arm, UNKNOWN)-String
health Health status of the node (OK, Warning, Critical)-String
inletTemperature Temperature of the inlet air°CDouble
outletTemperature Temperature of the outlet air°CDouble
highestTemperature Highest temperature measured on the node's baseboard°CDouble
voltage Supply voltage of the baseboardVDouble
lastSensorUpdate Timestamp of the last sensor updatemsLong
macAddressCompute MAC address of the NIC connected to the compute network (optional)-String
macAddressMgmt MAC address of the NIC connected to the management network (optional)-String

In accordance to the component node the API offers nodeList which returns multiple instances of node.

Backplane

Example XML:

<backplane rcuPosition="1" id="RCU_84055620466592_BP_1" infrastructurePower="0.0">
<temperatures>24.0</temperatures>
<temperatures>25.0</temperatures>
<temperatures>26.0</temperatures>
<temperatures>27.0</temperatures>
<temperatures>28.0</temperatures>
</backplane>

The attributes have the following meaning:

Attribute Description Unit Data type
id Unique ID for referencing the component-String
rcuPosition Position of the backplane in the RECS®|Box Computing Unit-Integer
infrastructurePower Power usage of the infrastructure components on the backplaneWDouble
temperatures List of temperatures measured on the backplane°CDouble

In accordance to the component backplane the API offers backplaneList which returns multiple instances of backplane.

Baseboard

Example XML:

<baseboard rcuPosition="6" baseboardType="APLS" id="RCU_84055620466592_BB_6" infrastructurePower="9.8" rcuId="RCU_84055620466592">
<nodeId>RCU_84055620466592_BB_6_1</nodeId>
<nodeId>RCU_84055620466592_BB_6_2</nodeId>
<nodeId>RCU_84055620466592_BB_6_3</nodeId>
<temperatures>20.0</temperatures>
<temperatures>20.0</temperatures>
<temperatures>20.0</temperatures>
<temperatures>20.0</temperatures>
<temperatures>20.0</temperatures>
</baseboard>

The attributes have the following meaning:

Attribute Description Unit Data type
id Unique ID for referencing the component-String
rcuId Unique ID of the RECS®|Box Computing Unit hosting the baseboard-String
rcuPosition Position of the baseboard inside the RECS®|Box Computing Unit-Integer
infrastructurePower Power usage of the infrastructure components on the baseboardWDouble
baseboardType Type of the baseboard (CXP, APLS)-String
nodeId List of IDs of the nodes installed on the baseboard-String
temperatures List of temperatures measured on the backplane°CDouble

In accordance to the component baseboard the API offers baseboardList which returns multiple instances of baseboard.

RCU

Example XML:

<rcu rcuType="ANTARES" fanSpeed="60" rackId="RCK_1" name="RECSMaster (RCU) on 192.168.56.195" rackPosition="0" id="RCU_84055620466592">
<backplaneId>RCU_84055620466592_BP_1</backplaneId>
<baseboardId>RCU_84055620466592_BB_1</baseboardId>
<baseboardId>RCU_84055620466592_BB_2</baseboardId>
<baseboardId>RCU_84055620466592_BB_3</baseboardId>
<baseboardId>RCU_84055620466592_BB_4</baseboardId>
<baseboardId>RCU_84055620466592_BB_5</baseboardId>
<baseboardId>RCU_84055620466592_BB_6</baseboardId>
</rcu>

The attributes have the following meaning:

Attribute Description Unit Data type
id Unique ID for referencing the component-String
rackId ID of the rack which hosts the RECS®|Box Computing Unit-String
rackPosition Position of the RECS®|Box Computing Unit in the rack-Integer
name Name of the RECS®|Box Computing Unit-String
ip IP address of the RECS®|Box Computing Unit-String
rcuType Type of the RECS®|Box Computing Unit (SIRIUS, ARNEB, ANTARES)-String
kvmNode ID of the node to which the KVM system is switched (optional)-String
fanSpeed Current speed setting of the fans in the RECS®|Box Computing Unit%Integer
backplaneId List of IDs of backplanes which are installed in the RECS®|Box Computing Unit-String
baseboardId List of IDs of baseboards which are installed in the RECS®|Box Computing Unit-String

In accordance to the component rcu the API offers rcuList which returns multiple instances of rcu.

Rack

Example XML:

<rack description="Default rack" id="RCK_1">
<rcuId>RCU_84055620466592</rcuId>
</rack>

The attributes have the following meaning:

Attribute Description Unit Data type
id Unique ID for referencing the component-String
description Description of the rack-String
rcuId List of IDs of RECS®|Box Computing Units which are installed in the rack-String

In accordance to the component rack the API offers rackList which returns multiple instances of rack.

Resources

The resources are split into monitoring resources (for pure information gathering) and management resources (for changing the system configuration or state).

Monitoring

For monitoring the following resources are available:

Attribute Description HTTP Method
/node Returns a nodeList with all nodes of the clusterGET
/node/{node_id} Returns information about the node with the given IDGET
/baseboard Returns a baseboardList with all baseboards of the clusterGET
/baseboard/{baseboard_id} Returns information about the baseboard with the given IDGET
/baseboard/{baseboard_id}/node Returns a nodeList with all nodes that are installed on the baseboard with the given IDGET
/backplane Returns a backplaneList with all backplanes of the clusterGET
/backplane/{backplane_id} Returns information about the backplane with the given IDGET
/rcu Returns an rcuList with all RECS®|Box Computing Units of the clusterGET
/rcu/{rcu_id} Returns information about the RECS®|Box Computing Unit with the given IDGET
/rcu/{rcu_id}/baseboard Returns a baseboardList with all baseboards that are installed in the RECS®|Box Computing Unit with the given IDGET
/rcu/{rcu_id}/backplane Returns a backplaneList with all backplanes that are installed in the RECS®|Box Computing Unit with the given IDGET
/rcu/{rcu_id}/node Returns a nodeList with all nodes that are installed in the RECS®|Box Computing Unit with the given IDGET
/rack Returns a rackList with all racks of the clusterGET
/rack/{rack_id} Returns information about the rack with the given IDGET
/rack/{rack_id}/rcu Returns a rcuList with all RECS®|Box Computing Units that are installed in the rack with the given IDGET

Management

The management of individual components can be found under the “manage” path of the component.

Attribute Description HTTP method Parameter
/node/{node_id}/manage/power_on Turns on the node with the given ID and returns updated node XMLPOST
/node/{node_id}/manage/power_off Turns off the node with the given ID and returns updated node XMLPOST
/node/{node_id}/manage/reset Resets the node with the given ID and returns updated node XMLPOST
/node/{node_id}/manage/select_kvm Switches the KVM port of the RECS®|Box Computing Unit containing the node to the node with the given ID and returns updated node XMLPUT
/rcu/{rcu_id}/manage/set_fans Sets the overall fan speed of the RCU with the given ID and returns the curent status of the RCUPUTpercent={value}

Errors

Information about the success or failure of management requests are returned via HTTP status codes. Please have a look at RFC2616 for an overview about the defined HTTP status codes.

Nagios API

The software integration work to monitor the RECS®|Box System is quite simple as the the TOR-Master provides monitoring information in the Nagios native NRPE format. So only the NRPE plugin has to be installed and configured as follows. Here, a sample output can be found:

$ /usr/lib/nagios/plugins/check_nrpe -H 10.11.12.244 \
            -c check_temp -a 10.11.12.244 10 2 70:104 105:
 
OK - Temperature: 29 C|temp=29,000000;70:104;105;70,000000;105,000000

The options are used as following:

Option Description
-H Host to ask for data, this is always the IP of the TOR-Master (example: 10.11.12.244)
-c Plugin to run. Can be check_temp or check_power
-a Arguments to pass to the plugin, see more details in tables below

Arguments for check_temp plugin:

Argument example Description
10.11.12.244 Get sensor values from device with this IP (RCU/RPU)
10 Get sensor values from this baseboard (1 - 18)
2 Get values from this sensor (max, inlet, outlet, 0, 1, 2, 3, 4)
70:104 Range of warning threshold
105 Range critical threshold (ending with : means open end)

Arguments for the check_power plugin:

Argument example Description
10.11.12.244 Get sensor values from device with this IP (RCU/RPU)
10 Get sensor values from this baseboard (1 - 18)
2 Get sensor values from this node (1, 2, 3, 4)
80:109 Range of warning threshold [Watt]
110: Range of critical threshold [Watt] (ending with : means open end