Gluster Health Report Tool

Use this tool to analyze the Gluster node for correctness or health.

Contributing

To contribute reports, please see CONTRIBUTING.

Usage

gluster-health-report

or run gluster-health-report --help for more details.

Install

Clone the repo and install using,

git clone https://github.com/aravindavk/gluster-health-report.git
cd gluster-health-report
sudo python setup.py install

How to add new report

Adding new report is very simple, add a python file in glusterhealth/reports/myreport.py and implement your report. Any function which starts with the name "report_" will be executed by the health checker framework.

Following example shows a report which checks Glusterd is running or not.

import logging
 
from utils import command_output, CommandError

def report_check_running(ctx):
    cmd = ["ps", "-C", "glusterd"]
    try:
        command_output(cmd)
        ctx.ok("Glusterd is running")
    except CommandError as e:
        ctx.notok("Glusterd is not running")
        logging.warn(ctx.lf("Glusterd is not running",
                            error_code=e[0],
                            error=e[1]))

To understand the report implementation, let us walk through line by line

Import logging library to use logging in your report. Note: Logging setup stuff is already taken care by the framework itself.

import logging

Import utility functions provided by the framework to execute the command.

from utils import command_output, CommandError

Implement your report

def report_check_running(ctx):

Executing ps command to check glusterd is running or not

    cmd = ["ps", "-C", "glusterd"]

If command returns zero/success, send message to framework using ctx.ok

    try:
        command_output(cmd)
        ctx.ok("Glusterd is running")

If command fails then send message to framework using ctx.notok. Also add entry to log file with additional details.

    except CommandError as e:
        ctx.notok("Glusterd is not running")
        logging.warn(ctx.lf("Glusterd is not running",
                            error_code=e[0],
                            error=e[1]))

Run Context

Each report function should accept context argument, which provides following methods.

ctx.lf - Format the log message in structured logging format
ctx.ok - Send a OK summary message back to framework
ctx.error - Send error summary message back to framework
ctx.notok - Send notok summary message back to framework
ctx.warning - Send warning summary message back to framework
ctx.rconf - Runtime configurations set by framework
    ctx.rconf.log_file - Log file path
    ctx.rconf.args - Parsed Arguments namespace
    ctx.rconf.enabled_reports - List of enabled reports, useful if
        any inter dependency

Dummy report

A dummy report is added, which just prints some messages without inspecting any health.

def report_hello(ctx):
    ctx.ok("CPU Usage")
    ctx.ok("Network Health")
    ctx.warning("Disconnect events")
    ctx.warning("Memory Usage")
    ctx.notok("Log rotate setup")
    ctx.ok("Error logs in last day")
    ctx.warning("Changelog size")

For testing it can be executed using

gluster-health-report --log-dir . --include hello

Output:

Loaded reports: hello
....
 
[     OK] CPU Usage
[     OK] Network Health
[WARNING] Disconnect events
[WARNING] Memory Usage
[ NOT OK] Log rotate setup
[     OK] Error logs in last day
[WARNING] Changelog size
 
....
You can find the detailed health-report at ./gluster-health-report-2017-09-20-16-54.log

TODO

• Execute report functions in parallel
• Add configuration for reports-dir
• Add more reports
• Add support for bash reports