docker-ce
and docker-compose
are installed on that machine. Installation instructions for installing Docker are on the Docker website.install.sh
script creates some directories and sets the permissions on them:export.yml
configuration file is used to configure weka-mon and the exporter. The export.yml
file can be found in the base of the weka-mon
directory hierarchy.cluster:
heading to reflect your hostnames or ip addresses; you need to specify one or more hostnames/ips - there's not need to list all the cluster hostnames; two or three will do.cluster:
is auth_token_file:
which is used to provide the security token required to authenticate with the cluster. This file can be generated with the weka user login
command on any cluster host (including clients) and copied to the server/VM running weka-mon. It is highly suggested that you create a ReadOnly User just for this package and use it for cluster communications. See the Security section in the Operations Guide for details on creating users and using tokens.exporter:
section that defines the program behavior.listen_port:
parameter defines the port that Prometheus should scrape. This should not be changed unless you change the Prometheus configuration.loki_host:
and loki_port:
parameters should not be changed if you're using the weka-mon setup. Make loki_host:
blank to disable sending events to Loki entirely.timeout:
parameter is the max time in seconds to wait for an API call to return. The default should be sufficient for most purposes.max_procs:
and max_threads_per_proc:
parameters define the scaling behavior. If the total number of hosts (servers and clients) exceeds max_threads_per_proc
, the exporter will spawn more processes accordingly. max_threads_per_proc
of 100, it would spawn 3 processes (280 / 100 = 2.8, round up to 3).max_procs
processes. If you have 1000's of hosts, it will double/triple up hosts on the threads.max_procs
of 8 and max_threads_per_proc
of 100, only 8 proccesses will be spawned, each with 100 threads, but there will be close to 4 hosts serviced per thread instead of the default 1 host per thread.export.yml
. To enable collecting the FILEATOMICOPEN_OPS statistic, remove the #
character at the beginning of the line. Note that if the statistic you wish to gather is in a Category that is commented out, you will need to uncomment the Category line as well if it is not already uncommented (the first line in the example below). Conversely, to stop collecting a statistic, comment out the statistic by inserting a #
at the beginning of the line.admin/admin
.JSON
files are in the subdirectory weka-mon/var_lib_grafana/dashboards
. Please follow the Grafana documentation on how to import the files.export.yml
configuration file. (see above)~/.weka directory
(so the container can read the auth file), /dev/log
so it can put entries in the Syslog, /etc/hosts
so it has some name resolution (you can also use DNS if your Docker environment is set up to do so), and finally mapping the config file (export.yml
) into the container.--help
or -h
for a full description.export-1.3.0.tar
file from the Version-1.3.0 release. Copy this file to your management server or VM. git clone https://github.com/weka/export
or go to https://github.com/weka/export/releases and download the source tarball.pip3 install -r requirements.txt
command to install all the required python modules.