This man page only documents the specific features of the dummy-ups driver. For information about the core driver, see nutupsdrv(8).
This program is a multi-purpose UPS emulation tool. Its general behavior depends on the running mode: "dummy" ("dummy-once" or "dummy-loop"), or "repeater".
In this mode, dummy-ups looks like a standard NUT device driver to upsd(8) and allows one to change any value for testing purposes. It is both interactive, controllable through the upsrw(1) and upscmd(1) commands (or equivalent graphical tool), and batchable through script files. It can be configured, launched and used as any other "real" NUT driver. This mode is mostly useful for development and testing purposes.
See below about the differences of dummy-once vs. dummy-loop modes — the former may be more suitable for "interactive" uses and tests.
In this mode, dummy-ups acts as a NUT client, simply forwarding data. This can be useful for supervision purposes. This mode can also allow some load sharing between several upsd instances communicating with ultimate NUT clients, with a "central" one using a point-to-point communication with the UPS. This arrangement can also help with networked UPSes, whose network management cards can be overwhelmed with a farm of servers directly polling SNMP or other protocols every few seconds.
The port specification in ups.conf depends on the running mode, and allows the driver to select the right mode of operation.
Since NUT v2.8.0, the mode specification in ups.conf allows users to override the mode of operation which would be otherwise guessed by the driver.
In this context, port in the ups.conf block defines a file name for the dummy-ups to read data from. This can either be an absolute or a relative path name. In the latter case the NUT sysconfig directory (i.e. /etc/nut, /usr/local/ups/etc, ...) is prepended.
Since NUT v2.8.0 two aspects of this mode are differentiated:
dummy-once reads the specified file once to the end (interrupting for TIMER lines, etc.) and does not re-process it until the filesystem timestamp of the data file is changed; this reduces run-time stress if you test with a lot of dummy devices, and allows use/test cases to upsrw variables into the driver instance — and they remain in memory until the driver is restarted (or the file is touched or modified);
Since NUT v2.8.0 dummy-once is assigned by default to files with a *.dev naming pattern.
dummy-loop reads the specified file again and again, with a short sleep between the processing cycles; for sequence files using a TIMER keyword (see below), or for use/test cases which modify file contents with external means, this allows an impression of a device whose state changes over time.
Before NUT v2.8.0 this was the only aspect, so a simple dummy mode value maps to this behavior for backwards compatibility.
Since NUT v2.8.0 dummy-loop is assigned by default to files with a *.seq naming pattern, and dummy is assigned by default to files with other naming patterns that the driver could not classify.
Said defaulting based on filename pattern can break third-party test scripts which earlier expected *.dev files to work as a looping sequence with a TIMER keywords to change values slowly. Now such files should get processed to the end once.
Specify mode=dummy-loop driver option or rename the data file used in the port option for legacy behavior.
Use/Test-cases which modified such files content externally should not be impacted.
[dummy1] driver = dummy-ups port = evolution500.seq desc = "dummy-ups in dummy-loop mode"
[dummy2] driver = dummy-ups port = epdu-managed.dev desc = "dummy-ups in dummy-once mode"
This file is generally named something.dev or something.seq. It contains a list of all valid variables and associated values (you can later use upsrw only to modify values of these variables), and has the same format as an upsc(8) dump (<varname>: <value>). So you can easily create definition files from an existing UPS using upsc > file.dev.
Note that the Network UPS project provides an extensive DDL (Devices Dumps Library) with files which can be used for modelling real devices. Entries for the DDL library are best prepared with the tools/nut-ddl-dump.sh script from NUT sources instead of plain upsc, to provide some additional data points from other NUT clients as well.
The file can also be empty, in which case only a basic set of data is available: device.*, driver.*, ups.mfr, ups.model, ups.status as filled by the driver itself.
Some sample definition files are available in the data directory of the NUT source tree, and generally in the sysconfig or share directory of your system distribution.
Since dummy-ups will usually loop on reading this file, you can dynamically modify it with some external process to "interact" with the driver. This will avoid message spam into your system log files, if you are using NUT default configuration.
By default since NUT v2.8.0, it will not loop on files in dummy-once mode, e.g. those with a .dev extension, unless their timestamp changes.
You can also use the TIMER <seconds> instruction to create scheduled event sequences (such files are traditionally named with the .seq extension). For example, the following sequence will loop on switching ups.status between "OL", "OB" and "OB LB" every minute:
ups.status: OL TIMER 60 ups.status: OB TIMER 60 ups.status: OB LB TIMER 60
It is wise to end the script for dummy-loop mode with a TIMER keyword. Otherwise dummy-ups will directly go back to the beginning of the file and, in particular, forget any values you could have just set with upsrw.
Note that to avoid CPU overload with an infinite loop, the driver "sleeps" a bit between file-reading cycles (currently this delay is hardcoded to one second), independently of (and/or in addition to) any TIMER keywords.
In this context, port in the ups.conf block is the name of a remote UPS, using the NUT format, i.e.:
[repeater] driver = dummy-ups port = ups1@remotehost desc = "dummy-ups in repeater mode"
Unlike UPS specifications in the rest of NUT, the @hostname portion is not optional - it is the @ character which enables Repeater Mode. To refer to an UPS on the same host as dummy-ups, use port = upsname@localhost.
Note that to avoid CPU overload with an infinite loop, the driver "sleeps" a bit between data-requesting cycles (currently this delay is hardcoded to one second), so propagation of data updates available to a remote upsd may lag by this much.
Once the driver is loaded in dummy mode, you can change any variables, except those of the driver.* and server.* collections. You can do this by either editing the definition file, or use the upsrw(1) and upscmd(1) commands.
Note that in simulation mode, new variables can be added on the fly, but only by adding these to the definition file (and waiting for it to be re-read). That is, the driver should not allow to define a new variable via upsrw.
Conversely, if you need to remove a variable (such as transient ones, like ups.alarm), simply update these by setting an empty value. As a result, they will get removed from the data.
In repeater mode, the driver acts according to the capabilities of the UPS, and so supports the same instant commands and settable values.
Dummy Mode was originally written in one evening to replace the previous dummycons testing driver, which was too limited, and required a terminal for interaction.
dummy-ups is useful for NUT client development, and other testing purposes.
It also helps the NUT Quality Assurance effort, by automating some tests on the NUT framework.
It now offers a repeater mode. This will help in building the Meta UPS approach, which allows one to build a virtual device, composed of several other devices (either UPS, PDUs), or perhaps represent the same device which supports several communication protocols and different media (Serial, USB, SNMP...)
Instant commands are not yet supported in Dummy Mode, and data need name/value checking enforcement, as well as boundaries or enumeration definition.
If you use service management frameworks like systemd or SMF to manage the dependencies between driver instances and the data server, and some of these drivers are dummy-ups in repeater mode representing data from another driver running on the same system, then you may have to set up special dependencies (e.g. with systemd "drop-in" snippet files) to allow your nut-server to start after the "real" device drivers and before such repeater drivers (without a responding server, they would fail to start anyway). This may also need special care in upsd.conf and/or ups.conf files to not block the system start-up for too long while the repeater driver has not started.
upscmd(1), upsrw(1), ups.conf(5), nutupsdrv(8)
The "repeater" mode of dummy-ups driver is in some ways similar to the clone driver, which sits on top of another driver socket, and allows users to group clients to a particular outlet of a device and deal with this output as if it were a normal UPS.
The NUT (Network UPS Tools) home page: http://www.networkupstools.org/
clone(8), nut-recorder(8), nutupsdrv(8), upsd(8).