Skip to content
GitLab

EAR commands · Changes

Page history
Update EAR commands authored by Lluis Alonso's avatar Lluis Alonso
Show whitespace changes
Inline Side-by-side
EAR-commands.md
View page @ 6b5a963c
EAR offers the following commands:
EAR offers a set of commands which help both users and administrators to interact with different components:
- Commands to analyze data stored in the DB: [eacct](#ear-job-accounting-eacct) and [ereport](#energy-report-ereport).
- Commands to control and temporally modify cluster settings: [econtrol](#energy-control-econtrol).
- Commands to create/update/clean the DB: [edb_create](#edb_create), [edb_clean_pm](#edb_clean_pm) and [edb_clean_apps](#edb_clean_apps).
- A command to run OpenMPI applications with EAR on SLURM systems through `mpirun` command: [erun](#erun).
- Commands to retrieve data stored in the DB: [eacct](#ear-job-accounting-eacct) and [ereport](#energy-report-ereport).
- Commands to control and temporary modify cluster settings: [econtrol](#energy-control-econtrol).
- Commands to create/update/clean the DB: [edb_create](#edb_create), [edb\_clean\_pm](#edb_clean_pm) and [edb\_clean\_apps](#edb_clean_apps).
- A command to load transparently the EAR Library on systems where the batch scheduler has not a plug-in nor some use case isn't supported (e.g., running an OpenMPI application on SLURM systems through the `mpirun` command): [erun](#erun).
- A command to show current EAR installation information: [ear-info](#ear-info).
Commands belonging to the first three categories read the EAR configurarion file
......@@ -18,23 +18,21 @@ Some options are disabled when the user is not authorized.
# EAR job Accounting (eacct)
The `eacct` command shows accounting information stored in the EAR DB for jobs
(and step) IDs.
The command uses EAR's configuration file to determine if the user running it is
privileged or not, as **non-privileged users can only access their information**.
The `eacct` is a simple command to see a jobs' energy accounting information.
It can also retrieve EARL events that occurred on a job execution.
The command uses the EAR Configuration file to determine whether the user running it is
authorized, as **non-privileged users can only access their information**.
It provides the following options.
```
Usage: eacct [Optional parameters]
Optional parameters:
-h displays this message
-v displays current EAR version
-b verbose mode for debugging purposes
-u specifies the user whose applications will be retrieved. Only available to privileged users. [default: all users]
-j specifies the job id and step id to retrieve with the format [jobid.stepid] or the format [jobid1,jobid2,...,jobid_n].
A user can only retrieve its own jobs unless said user is privileged. [default: all jobs]
-a specifies the application names that will be retrieved. [default: all app_ids]
-c specifies the file where the output will be stored in CSV format. If the argument is "no_file" the output will be printed to STDOUT [default: off]
-c specifies the file where the output will be stored in CSV format. [default: no file]
-t specifies the energy_tag of the jobs that will be retrieved. [default: all tags].
-s specifies the minimum start time of the jobs that will be retrieved in YYYY-MM-DD. [default: no filter].
-e specifies the maximum end time of the jobs that will be retrieved in YYYY-MM-DD. [default: no filter].
......@@ -45,6 +43,7 @@ Usage: eacct [Optional parameters]
-o modifies the -r option to also show the corresponding jobs. Should be used with -j.
-n specifies the number of jobs to be shown, starting from the most recent one. [default: 20][to get all jobs use -n all]
-f specifies the file where the user-database can be found. If this option is used, the information will be read from the file and not the database.
-b verbose mode for debugging purposes
```
The basic usage of `eacct` retrieves the last 20 applications (by default) of the
......@@ -259,6 +258,9 @@ Usage: econtrol [options]
--restore-conf ->restores the configuration for all nodes
--active-only ->supresses inactive nodes from the output in hardware status.
--health-check ->checks all EARDs and EARDBDs for errors and prints all that are unresponsive.
--domain [domain:target] ->sends the requested command to the requested targets, effectively filtering
which nodes receive the message. Available domains are: tag, node, subcluster/eargmid, island.
--mail [address] ->sends the output of the program to address.
--ping ->pings all nodes to check whether the nodes are up or not. Additionally,
--ping=node_name pings that node individually.
......@@ -291,6 +293,65 @@ Node id Job-Step M-Rank DC power CPI GBS Gflops Time Avg
node2 6878-0 0 280.13 0.37 24.39 137.57 56.00 2.59
node3 6878-0 1 245.44 0.37 24.29 136.40 56.00 2.59
```
A list of nodes can be specified to only target those for the commands:
```
[user@login]$ econtrol --status --hosts node2,node3
hostname power temp freq job_id stepid
node2 278 66C 2.59 6878 0
node3 274 57C 2.59 6878 0
[user@login]$ econtrol --status --hosts island[0,1]node[2-3]
hostname power temp freq job_id stepid
island0node2 278 66C 2.59 0 0
island0node3 274 57C 2.59 0 0
island1node2 273 56C 2.59 0 0
island1node3 272 57C 2.59 0 0
```
If only one node is targeted for a status, one may do:
```
[user@login]$ econtrol --status=island0node2
hostname power temp freq job_id stepid
island0node2 278 66C 2.59 0 0
```
For any other command type (including status):
```
[user@login]$ econtrol --status --hosts island0node2
hostname power temp freq job_id stepid
island0node2 278 66C 2.59 0 0
[user@login]$ econtrol --restore-conf --hosts island0node2
[user@login]$ econtrol --status --domain node:island0node2
hostname power temp freq job_id stepid
island0node2 278 66C 2.59 0 0
```
Furthermore, the domain option may be used to filter out any nodes not belonging to a specified domain.
To only send a command to a single island (as defined in ear.conf):
```
[user@login]$ econtrol --restore-conf --domain island:2
```
To send to all the nodes that belong to a tag (as defined in ear.conf), regardless of their island or any other configuration:
```
[user@login]$ econtrol --set-powercap 400 --domain tag:cpu_only
```
One can also do use a double filter:
```
[user@login]$ econtrol --set-freq 3100000 --domain tag:cpu_only island:3
```
**NOTE**: Using domain filtering with a hostlist specification is not supported and may cause some errors.
# Database commands
......