Add a new request handler for the root URL (`/`) to render a simple HTML
page with a link to the metrics page. This follows the convention of other
metric exporters.
Update the exporter startup to add more log messages. The server address,
port, and metrics path are now logged on startup.
A "ready" log message is printed when the server is up and running.
Split out all the code to define exporter functions and collect data into
a new package. The new package is responsible for all exporter related
activity. This makes the code easier to read.
Split the code for collecting metrics from the database and from the socket
into different files to make the separation more obvious.
Add a new `f2b_version` metric that includes the version of the fail2ban
server and the exporter.
Add a new socket command to get back the fail2ban server version.
Update the project changelog.
Refactor the changelog structure to no longer be fully automated. It now
includes some manual tweaks.
Update the Makefile command to generate a new `CHANGELOG_gen.md` file
instead of overwriting the existing Changelog file.
Update all old database-based metrics to include the `deprecated` text.
Add a warning on startup if connecting to the fail2ban database to state
that this functionality will be removed in a future release.
Rename deprecated methods and variables.
Add new metric to collect the number of errors found when connecting to the
fail2ban server socket. Errors are split into two categories: connection
errors (e.g. socket file not found), and request errors (e.g. invalid
response received from server).
Update the `up` metric to return `0` if the socket connection fails.
Improve error logging.
Update the code collecting metrics to open a new socket connection each
time metrics are collected. This ensures that a new socket connection is
used each time and avoids errors caused by fail2ban being restarted.
Update the docker container to support mounting the fail2ban server socket
and pointing the exporter at it. This allows the exporter to interact with
the socket from within the container.
The entire `/var/run` folder is mounted instead of just the socket file to
correctly handle fail2ban restarts (where the file will be deleted).
Add new metric to track the total number of jails configured in fail2ban.
Add new metrics for the current and total number of filter failures for
each jail, as well as the current/total number of banned IPs per jail.
The new metrics are collected by sending the `status [jail]` command to the
fail2ban server and parsing the response data.
Read the response data from the socket in chunks to prevent errors when
processing large payloads. The initial implementation solved large payloads
by just defining a very large buffer, but this is not a solution. The new
code reads the socket data in a loop until a terminator is found and
appends all the data into a single byte array.
Reduce the buffer size to `1024` bytes.
Add support for connecting the exporter directly to the fail2ban server's
socket to send requests and receive data. The path to the socket file is
optional and specified on startup.
Export a new metric based on the response of the `ping` command sent to the
fail2ban server. The metric is set to 1 if the server responds with `pong`
and 0 in any other case. This metric is only shown if the path to the
socket file was provided on startup.
Update the database query counting the number of banned IPs to filter out
any bans that have already expired. An expired ban is defined as a ban
where the "time of ban" plus the "duration of ban" is less than the
current time.
This is necessary because bans that have expired are not automatically
removed from the database and will cause metrics to diverge from the counts
reported by `fail2ban-client`.
Add a new metric to count the number of database errors that have been
found since startup. This complements the `up` metric to have better
visibility into occasional database errors.
The `up` metric is now based on whether an error was found while reading
data from the database to build other metrics. Note that there is a chance
the `up` metric will not be correctly set if the last metric to be built
before the `up` metric does not throw an error.
Add a new prometheus metric to track which jails are currently enabled.
Add a new database query to read the jail name and enabled status from the
database.
Add new metric to readme file.
