HMCi is a utility that collects metrics from one or more IBM Power Hardware Management Consoles (HMC), without the need to install agents on logical partitions / virtual machines running on the IBM Power systems. The metric data is processed and saved into an InfluxDB time-series database. Grafana is used to visualize the metrics data from InfluxDB through provided dashboards, or your own customized dashboards.
Go to file
Mark Nellemann 2c1921564b
continuous-integration/drone/push Build is passing Details
continuous-integration/drone/tag Build is passing Details
Merge pull request 'Avoid HMC sessions timeouts.' (#2) from sessiontimeouts into main
Reviewed-on: #2
2023-11-13 13:16:45 +00:00
doc Work on avoiding lingering sessions on the HMC. 2023-08-10 11:02:53 +02:00
gradle/wrapper Update gradle and fix typo 2023-01-05 14:32:57 +01:00
src Update 3rd party dependencies. 2023-11-13 14:14:27 +01:00
.drone.yml Preperations for migration. 2023-01-04 15:41:59 +01:00
.editorconfig Fix data-source in dashboards. 2021-06-04 08:27:16 +02:00
.gitattributes Initial code commit 2020-08-07 08:13:48 +02:00
.gitignore Cleanup code. 2020-12-01 11:48:05 +01:00
CHANGELOG.md Update 3rd party dependencies. 2023-11-13 14:14:27 +01:00
LICENSE Added LICENSE information 2020-08-18 13:49:48 +02:00
README.md Updated dashboards, docs and 3rd party deps. 2023-05-19 20:37:24 +02:00
bitbucket-pipelines.yml Major refactoring of xml+json deserialization. 2022-11-24 12:35:49 +01:00
build.gradle Update 3rd party dependencies. 2023-11-13 14:14:27 +01:00
gradle.properties Work on avoiding lingering sessions on the HMC. 2023-08-10 11:02:53 +02:00
gradlew Update gradle and fix typo 2023-01-05 14:32:57 +01:00
gradlew.bat Update gradle and fix typo 2023-01-05 14:32:57 +01:00
settings.gradle Update 3rd party deps., and implement option disable query for energy metrics 2022-02-08 21:39:42 +01:00

README.md

HMC Insights / HMCi

HMCi is a utility that collects metrics from one or more IBM Power Hardware Management Consoles (HMC), without the need to install agents on logical partitions / virtual machines running on the IBM Power systems. The metric data is processed and saved into an InfluxDB time-series database. Grafana is used to visualize the metrics data from InfluxDB through provided dashboards, or your own customized dashboards.

This software is free to use and is licensed under the Apache 2.0 License, but is not supported or endorsed by International Business Machines (IBM).

Metrics includes:

  • Managed Systems - the physical Power servers
  • Logical Partitions - the virtualized servers running AIX, Linux and/or IBM-i (AS/400)
  • Virtual I/O Servers - the i/o partition(s) virtualizing network and storage
  • Energy - watts and temperatures (needs to be enabled and is not available on all systems)

architecture

Some of my other related projects are:

  • svci for monitoring IBM Spectrum Virtualize (Flashsystems / Storwize / SVC)
  • sysmon for monitoring all types of servers with a small Java agent
  • syslogd for redirecting syslog and GELF to remote logging destinations

Screenshots

screenshot

Screenshots of other dashboards are available in the screenshots folder.

Installation and Setup

There are few steps in the installation.

  1. Preparations of the Hardware Management Console (HMC)
  2. Installation of InfluxDB, Grafana and the HMCi (this) software
  3. Configure Grafana and import example dashboards

1 - Preparations of the Hardware Management Console (HMC)

On the HMC we create a viewer user which are used to fetch data through it's REST API and enable collection of performance metrics.

Read the readme-hmc.md file for instructions and help.

2 - Installation of InfluxDB, Grafana and the HMCi software

The required software can be installed together on the same LPAR / VM or separately, as long as it's possible to ensure network connectivity.

We are going to install HMCi (this software) to fetch and process metrics from our HMC, InfluxDB (v. 1.8 or 1.9) for storing the metrics data and lastly Grafana for visualizing the data.

Depending on your operating system, go through one of the documents:

3. Configure Grafana and import example dashboards

Read the readme-grafana.md file for instructions and help.

Notes

No data (or past/future data) shown in Grafana

This is most likely due to timezone, date and/or NTP not being configured correctly on the HMC and/or host running HMCi.

You can check the timestamp of the most recent data by querying InfluxDB with the influx CLI client, and take note of the timezone when comparing:

use hmci;
precision rfc3339;
SELECT * FROM server_details GROUP BY * ORDER BY DESC LIMIT 1;

Compatibility with nextract Plus

From version 1.2 HMCi is made compatible with the similar nextract Plus tool from Nigel Griffiths. This means that the Grafana dashboards made by Nigel are compatible with HMCi and the other way around.

InfluxDB Retention Policy

Examples for changing the default InfluxDB retention policy for the hmci database:

ALTER RETENTION POLICY "autogen" ON "hmci" DURATION 156w
ALTER RETENTION POLICY "autogen" ON "hmci" DURATION 90d

Upgrading HMCi

On RPM based systems (RedHat, Suse, CentOS), download the latest hmci-x.y.z-n.noarch.rpm file and upgrade:

rpm -Uvh hmci-x.y.z-n.noarch.rpm

On DEB based systems (Debian, Ubuntu and derivatives), download the latest hmci_x.y.z-n_all.deb file and upgrade:

dpkg -i hmci_x.y.z-n_all.deb

Restart the HMCi service on systemd based Linux systems:

systemctl restart hmci
journalctl -f -u hmci  # to check log output

On AIX you need to pass the --ignoreos flag to the rpm command:

rpm -Uvh --ignoreos hmci-x.y.z-n.noarch.rpm

Possible problems

Incomplete set of metrics

I have not been able to test and verify all types of metric data. If you encounter any missing or wrong data, please contact me and I will try to fix it.

Naming collision

You can't have partitions (or Virtual I/O Servers) on different Systems with the same name, as these cannot be distinguished when metrics are written to InfluxDB (which uses the name as key).

Renaming partitions

If you rename a partition, the metrics in InfluxDB will still be available by the old name, and new metrics will be available by the new name of the partition. There is no easy way to migrate the old data, but you can delete it easily:

DELETE WHERE lparname = 'name';

Development Information

You need Java (JDK) version 8 or later to build hmci.

Build & Test

Use the gradle build tool, which will download all required dependencies:

./gradlew clean build

Local Testing

InfluxDB v1.x

Start a InfluxDB container:

docker run --name=influxdb --rm -d -p 8086:8086 influxdb:1.8

Create the hmci database:

docker exec -i influxdb influx -execute "CREATE DATABASE hmci"

InfluxDB v2.x

Start a InfluxDB container:

docker pull influxdb:latest
docker run --name=influxdb --rm -d -p 8086:8086 influxdb:latest
  • Then use the Web UI to create an initial user (for the web UI), an organization and bucket: http://localhost:8086/
  • Then create an API token with RW access to your bucket.

Grafana

Start a Grafana container, linking it to the InfluxDB container:

docker run --name grafana --link influxdb:influxdb --rm -d -p 3000:3000 grafana/grafana

Setup Grafana to connect to the InfluxDB container by defining a new datasource on URL http://influxdb:8086 named hmci.

If you are connecting to InfluxDB v2.x, then add a custom http header, enter bucket as database and disable authorization.

  • Authorization = Token abcdef_random_token_from_nfluxdb==

Import dashboards from the doc/dashboards/ folder.