Skip to content

Latest commit

 

History

History
117 lines (95 loc) · 6.6 KB

README.md

File metadata and controls

117 lines (95 loc) · 6.6 KB
Raw

Hardware Management Collector (hmcollector)

The hmcollector listens for power on and off events and collects telemetry data.

How it Works

There are two pods cray-hms-hmcollector-ingress and cray-hms-hmcollector-poll.

At start up cray-hms-hmcollector-poll creates redfish subscriptions on the BMCs. It gets the the list of BMCs from HSM's redfish endpoints (/hsm/v2/Inventory/RedfishEndpoints). After setting up the subscriptions it starts polling telemetry data via redfish for the hardware that does not support telemetry subscriptions.

The cray-hms-hmcollector-ingress acts as an HTTP server that the BMCs make HTTP POST requests to when the BMC has some information to report. When this call back happens hmcollector parses the request and puts the information on the kafka bus for other CSM services to use. For example, the HSM services uses this information to change the State field on its Components data.

Configuration

To edit one of the configuration variables on a deployed hmcollector run the following commands

kubectl -n services edit deployment/cray-hms-hmcollector-ingress
kubectl -n services edit deployment/cray-hms-hmcollector-poll

Then make changes to the given variable and save and exit. The given pods will restart if changes have been made. These changes will last until the hmcollector is upgraded.

Environment Variable Description
HMCOLLECTOR_CA_URI URI of the CA cert bundle
HSM_REFRESH_INTERVAL The interval to check HSM for new Redfish Endpoints in seconds
LOG_LEVEL Sets the log level: ERROR, INFO, DEBUG
LOG_MODES Custom logging options. Supported values: errors
LOG_XNAMES Enable logging of ingress messages for the given BMC xnames. For example: "x9000c1s2b3,x9000c1s2b4" (Note that spaces are not allowed)
POLLING_ENABLED Enables or disables polling
POLLING_INTERVAL The polling interval in seconds for redfish telemetry polling
PRUNE_OLD_SUBSCRIPTIONS Enable pruning old subscriptions that contain the wrong xname for the given BMC
REST_ENABLED Controls whether or not the HTTP server is started
REST_URL The Address for Redfish events to target
RF_STREAMING_ENABLED Enable creating subscriptions for telemetry
RF_SUBSCRIBE_ENABLED Enable creating redfish subscriptions
SM_URL The address of the State Manager (aka hsm, aka smd)
VAULT_ADDR The address of the vault server
VAULT_ENABLED Enable using vault for credentials
VAULT_KEYPATH Keypath for Vault credentials

Redfish Examples

Get a list of the subscriptions

curl -ks -X GET -u root:${BMC_PASSWD} https://${XNAME}/redfish/v1/EventService/Subscriptions | jq

Get subscription details (example from Mountain hardware)

curl -ks -X GET -u root:${BMC_PASSWD} https://${XNAME}/redfish/v1/EventService/Subscriptions/1 | jq

Delete a subscription

curl -ksi -X DELETE -u root:${BMC_PASSWD} https://${XNAME}/redfish/v1/EventService/Subscriptions/1

Create a subscription for power events

curl -ksi -X POST -H "Content-Type: application/json" \
    -u root:${BMC_PASSWD} \
    -d '{"Destination":"http://10.94.200.71/'${XNAME}'", "Protocol": "Redfish", "EventTypes": ["Alert", "StatusChange", "ResourceUpdated", "ResourceAdded", "ResourceRemoved"], "Context":"'${XNAME}'" }' \
    https://${XNAME}/redfish/v1/EventService/Subscriptions

Some hardware only support the Alert event type.

curl -ksi -X POST -H "Content-Type: application/json" \
    -u root:${BMC_PASSWD} \
    -d '{"Destination":"http://10.94.200.71/'${XNAME}'", "Protocol": "Redfish", "EventTypes": ["Alert"], "Context":"'${XNAME}'" }' \
    https://${XNAME}/redfish/v1/EventService/Subscriptions

Create a subscription for telemetry information

curl -ksi -X POST -H "Content-Type: application/json" \
    -u root:${BMC_PASSWD} \
    -d '{"Destination":"http://10.94.200.71/'${XNAME}'", "Protocol": "Redfish", "EventTypes": ["Alert", "StatusChange", "ResourceUpdated", "ResourceAdded", "ResourceRemoved"], "Context":"'${XNAME}'", "RegistryPrefixes": ["CrayTelemetry"] }' \
    https://${XNAME}/redfish/v1/EventService/Subscriptions

curl -ksi -X POST -H "Content-Type: application/json" \
    -u root:${BMC_PASSWD} \
    -d '{"Destination":"http://10.94.200.71/'${XNAME}'", "Protocol": "Redfish", "Context":"'${XNAME}'", "RegistryPrefixes": ["CrayTelemetry"] }' \
    https://${XNAME}/redfish/v1/EventService/Subscriptions

kubernetes actions

Restart the hmcollector

kubectl -n services rollout restart deployment cray-hms-hmcollector-ingress
kubectl -n services rollout restart deployment cray-hms-hmcollector-poll

Scale out the hmcollector ingress service from 3 pods to 6 pods. There needs to be one worker available for each pod.

kubectl -n services scale deployment cray-hms-hmcollector-ingress --current-replicas=3 --replicas=6

CT Testing

CT tests have not yet been written for hmcollector.