|
1 | 1 | # Running a Pbench Agent-driven Workload in a Container |
2 | | -1. Create a workload script (`example-workload.sh` can be used to start) |
3 | | - This script performs the invocation of a Pbench Agent workload driver, e.g. |
4 | | - `pbench-user-benchmark`, or `pbench-fio`, or `pbench-uperf`. |
5 | | -2. Setup the local execution environment to be aware of the target Pbench Server |
6 | | - `./setup.sh <pbench server host name>:<port number>` |
7 | | -3. Execute the workload |
8 | | - Typically this script performs the necessary setup required for the |
9 | | - containerized environment (`example-driver.sh` can be used to start, note |
10 | | - that it also moves the results to the Pbench Server). |
| 2 | + |
| 3 | +The Pbench Agent is available for use in a zero-installation scenario |
| 4 | +via containerized execution. Agent containers are available from |
| 5 | +[quay.io/pbench](https://quay.io/organization/pbench): there's a |
| 6 | +repository for each distribution (e.g., |
| 7 | +[pbench-agent-all-centos-9](https://quay.io/repository/pbench/pbench-agent-all-centos-9?tab=tags)) |
| 8 | +with tags for each Agent release as well as for "hot builds" for |
| 9 | +each development branch. |
| 10 | + |
| 11 | +The `pbench` script provided here is a wrapper which facilitates |
| 12 | +the invocation of a Pbench Agent command using a containerized |
| 13 | +deployment of the Pbench Agent. Simply prefix a Pbench Agent |
| 14 | +command line with the path to this script to run it inside a |
| 15 | +container, without needing to install the Agent on the host |
| 16 | +system. (This is easily done by defining a shell alias for it.) |
| 17 | + |
| 18 | +The provided `pbench_demo` script shows the sequence of commands |
| 19 | +which might be used to perform a `fio` benchmark run. |
0 commit comments