networkupstools / nut Goto Github PK

The idea behind the PowerChain is simple:
There are many links in a powerchain, that supplies power to a device (server, appliance, ...):

• power supply unit,
• UPS,
• PDU.
• STS (static transfer switch),
• bypass module,
• genset,
• ...

Monitoring only UPS is not sufficient to guaranty a smart power protection.
Having a consolidated view on the whole also allows real HA, Green Grid and SmartGrid features.

An example PowerChain for the CLC would be:

PSU1 ==> PDU1:outlet1 ==> UPS1 ==> Main Power
PSU2 ==> PDU2:outlet1 ==> UPS2 ==> Main Power

Other details:

• a parent can have multiple children (UPS, PDU, ...)

• A child can have multiple parents (STS)

  The branch "powerchain" implements this feature: https://github.com/networkupstools/nut/tree/powerchain

Tasks list:

• Create a new power supply unit (PSU) driver,

• Create a new NUT data, 'device.parent', defined in ups.conf and exposed by all NUT drivers.
  This will store the parent reference in NUT canonical form:

  device[:outlet][@hostname[:port]]

  Example:
  ups.conf
  [psu1]
  driver = nut-psu
  port = psu1
  parent = pdu1:outlet1@localhost
  [pdu1]
  driver = snmp-ups
  port =
  parent = ups1@localhost
  [ups1]
  driver = usbhid-ups
  port = auto
  parent = main

  $ upsc psu1 device.parent
  pdu1:outlet1@localhost

• Implement power-chain support in upsc ('-P' option to get a tree list, when using '-l' and '-L')
  Note: only the first version, with coma separators.

  $ upsc -P localhost
  psu1 -> pdu1:outlet1 -> ups1

  $ upsc -Pl localhost
  ups1
  |-> pdu1:outlet1
  |-> pdu1

• Implement a *tree store for upsd, to fix the lack of the current one (powerchain_t, in server/powerchain.ch

• Implement power-chain support in upsmon

There is no registry key to point NUT installation path, as selected during installation.
The right and suitable key name is to be confirmed by a Windows guru, but something like the following should probably do it:
HKEY_LOCAL_MACHINE\SOFTWARE\Network UPS Tools\InstallationPath

Checklist for the 2.7.1 interim release:

• @clepple: update dowload info for git
• update client info with walNUT
• address issue #30 (NSS support broke upsc output)
• address issue #15 (Avoid "2.6.x-Unversioned Directory" version numbers)
• check branch bcmxcp and merge if ok
• check branch voltronic-driver and merge if ok (but see #57)
• #57: check branch blzr and merge if ok; possibly removing voltronic
• add Tripp Lite information to drivers.list
• check branch powercom-instant-commands and merge if ok
• update news/upgrading files

Hello,

I reported a problem with the USB driver and an Eaton 5PX UPS. usbhid-ups failed to start, because it timeouts while receiving the descriptor. The USB_TIMEOUT constant in libusb.c is set to 4 seconds, instead of the standard 5 seconds.

Here is the original thread: http://www.mail-archive.com/[email protected]/msg07676.html

Here is the constant I'm talking about: https://github.com/networkupstools/nut/blob/master/drivers/libusb.c#L38

• Alioth patch #313714
  https://alioth.debian.org/tracker/index.php?func=detail&aid=313714&group_id=30602&atid=411544
• Available on branch Vaclav
• this has superseded Alioth bug #313634, Use difftime for time comparison:
  https://alioth.debian.org/tracker/index.php?func=detail&aid=313634&group_id=30602&atid=411542

With the apcupsd-ups driver merged into master, it would be handy to be able to scan for existing apcupsd NIS servers.

Basic detection would be an open TCP port 3551.

Refer to issue #23

This should address both usbhid-ups and mge-shut, since they share most of the code and logic. Possibly through a generic "com-timeout" option

Legacy work (refer to ticket #101):

• Alioth patch ##312563: https://alioth.debian.org/tracker/index.php?func=detail&aid=312563&group_id=30602&atid=411544
• this version was limited to RFC1628, thus 1 device of type UPS

A more recent effort, from @luizluca (Luiz Angelo Daros de Luca) is providing an agent and a MIB that provide NUT features (multiple devices, of various types):
https://github.com/luizluca/nut-snmpagent

See discussion:

• General: http://thread.gmane.org/gmane.comp.monitoring.nut.user/8218
• Specific to upsmon: http://thread.gmane.org/gmane.comp.monitoring.nut.user/8204
• upsd/driver events: http://article.gmane.org/gmane.comp.monitoring.nut.user/8233
• getaddrinfo(3) in upsd [citation needed]
• instcmd() in usbhid-ups: http://thread.gmane.org/gmane.comp.monitoring.nut.user/8528/focus=8533

Split jNUT source code and distribution from the mainline.
Create a new repository on github, dedicated to jNUT.

Create the 4th generation driver to address all older Q1 and newer Q5 / QFS protocol variations.

The general idea is to have something between the current blazer* and usbhid-ups like claiming and data mapping structure.

A special care must be taken to old dumb things:
current parameters have to be kept, and a "force q1" should probably be added.

Hi,
I discovered an issue when trying to configure an APC Back-UPS ES 700G.

As I assume from the documentation nut-scanner -qNU should create a usable configuration entry which can be directly copied into ups.conf.

When running the above command I get:

[nutdev1]
        driver = "usbhid-ups"
        port = "auto"
        vendorid = "051D"
        productid = "0002"
        product = "Back-UPS ES 700G FW:871.O2 .I USB FW:O2 "
        serial = "5B1243T01934  "
        vendor = "APC"
        bus = "002"

For a first quick test I used:
usbhid-ups -DD -u root -a nutdev1
------------%<----------------------------------------------------------------------------------------

Network UPS Tools - Generic HID driver 0.37 (2.6.5)
USB communication driver 0.32
   0.000000     debug level is '2'
   0.001513     upsdrv_initups...
   0.083102     Checking device (8087/0024) (001/002)
   0.083195     - VendorID: 8087
   0.083211     - ProductID: 0024
   0.083225     - Manufacturer: unknown
   0.083239     - Product: unknown
   0.083253     - Serial Number: unknown
   0.083267     - Bus: 001
   0.083281     Trying to match device
   0.083306     Device does not match - skipping
   0.083347     Checking device (1D6B/0002) (001/001)
   0.084525     - VendorID: 1d6b
   0.084542     - ProductID: 0002
   0.084556     - Manufacturer: Linux 3.4.41-dist ehci_hcd
   0.084570     - Product: EHCI Host Controller
   0.084584     - Serial Number: 0000:00:1a.0
   0.084598     - Bus: 001
   0.084612     Trying to match device
   0.084627     Device does not match - skipping
   0.084666     Checking device (8087/0024) (002/002)
   0.084740     - VendorID: 8087
   0.084755     - ProductID: 0024
   0.084769     - Manufacturer: unknown
   0.084783     - Product: unknown
   0.084797     - Serial Number: unknown
   0.084811     - Bus: 002
   0.084824     Trying to match device
   0.084839     Device does not match - skipping
   0.084877     Checking device (1D6B/0002) (002/001)
   0.086036     - VendorID: 1d6b
   0.086052     - ProductID: 0002
   0.086066     - Manufacturer: Linux 3.4.41-dist ehci_hcd
   0.086080     - Product: EHCI Host Controller
   0.086094     - Serial Number: 0000:00:1d.0
   0.086108     - Bus: 002
   0.086122     Trying to match device
   0.086137     Device does not match - skipping
   0.086175     Checking device (051D/0002) (002/004)
   0.117261     - VendorID: 051d
   0.117312     - ProductID: 0002
   0.117365     - Manufacturer: APC
   0.117411     - Product: Back-UPS ES 700G FW:871.O2 .I USB FW:O2 
   0.117452     - Serial Number: 5B1243T01934  
   0.117482     - Bus: 002
   0.117529     Trying to match device
   0.117660     Device does not match - skipping
   0.117743     No appropriate HID device found
   0.117799     No matching HID UPS found

------------%<----------------------------------------------------------------------------------------

So strangely nut-scanner is detecting the UPS but the driver is not detecting it.

First I tried to cherry-pick some fixes from the git master, applying them when building from the latest release source tarball (2.6.5).

• 3a89af5 - [nut-scanner] Fix a crash when no start IP is provided.

• 9758dcf - Set USB timeout to 5 seconds

but both patches didn't changed the behavior.

After that I simply tried to uncomment entries from the generated configuration and found out that when uncommenting the product and serial entry, the driver is able to detect the UPS. Same applies when only removing the trailing spaces from product and serial entry.

usbhid-ups -DD -u root -a nutdev1
------------%<----------------------------------------------------------------------------------------

Network UPS Tools - Generic HID driver 0.37 (2.6.5)
USB communication driver 0.32
   0.000000     debug level is '2'
   0.001530     upsdrv_initups...
   0.083640     Checking device (8087/0024) (001/002)
   0.083731     - VendorID: 8087
   0.083746     - ProductID: 0024
   0.083761     - Manufacturer: unknown
   0.083775     - Product: unknown
   0.083789     - Serial Number: unknown
   0.083803     - Bus: 001
   0.083817     Trying to match device
   0.083842     Device does not match - skipping
   0.083884     Checking device (1D6B/0002) (001/001)
   0.085067     - VendorID: 1d6b
   0.085083     - ProductID: 0002
   0.085098     - Manufacturer: Linux 3.4.41-dist ehci_hcd
   0.085112     - Product: EHCI Host Controller
   0.085126     - Serial Number: 0000:00:1a.0
   0.085140     - Bus: 001
   0.085154     Trying to match device
   0.085168     Device does not match - skipping
   0.085207     Checking device (8087/0024) (002/002)
   0.085282     - VendorID: 8087
   0.085297     - ProductID: 0024
   0.085312     - Manufacturer: unknown
   0.085326     - Product: unknown
   0.085339     - Serial Number: unknown
   0.085353     - Bus: 002
   0.085368     Trying to match device
   0.085382     Device does not match - skipping
   0.085420     Checking device (1D6B/0002) (002/001)
   0.086586     - VendorID: 1d6b
   0.086602     - ProductID: 0002
   0.086616     - Manufacturer: Linux 3.4.41-dist ehci_hcd
   0.086630     - Product: EHCI Host Controller
   0.086644     - Serial Number: 0000:00:1d.0
   0.086658     - Bus: 002
   0.086672     Trying to match device
   0.086687     Device does not match - skipping
   0.086725     Checking device (051D/0002) (002/004)
   0.118688     - VendorID: 051d
   0.118711     - ProductID: 0002
   0.118726     - Manufacturer: APC
   0.118740     - Product: Back-UPS ES 700G FW:871.O2 .I USB FW:O2 
   0.118754     - Serial Number: 5B1243T01934  
   0.118768     - Bus: 002
   0.118782     Trying to match device
   0.118887     Device matches
   0.125217     HID descriptor length 929
   0.272566     Report Descriptor size = 929
   0.272777     Using subdriver: APC HID 0.95

------------%<----------------------------------------------------------------------------------------

As we can see the UPS is detected when removing the trailing spaces from the affected config entries.

dummycons (1rst generation) and dummy-ups (2nd generation) have well served various purposes, such as:

• debug of the NUT framework,
• client development
• simulation for QA regression testing purpose (with static definitions and power sequences)

The work on dummy-ups also permitted the creation of the NUT Devices Dumps Library (http://www.networkupstools.org/devdumps/).

However, time has come for a 3rd generation to address the following, either in definition files (.dev / . seq) and / or the driver:

• 1. Add versioning to current definition files (.dev / . seq)
• 2. Specify v2 of definition files (.nut?), including the below
• 2.1. Handle read/write flags for variables
• 2.2. Handle instant commands
• 2.3. Handle complementary information and user comments (need specific comment annotation)
• 3. Create a new nutdrv-repeater driver from the "repetition" mode code
• 4. Create a new driver nutdrv-simulation from the "dummy" mode code
• 5. Create a tool to assist users in dumping all needed data in a smart and easy way

1. and 2.* will also permit to improve the NUT Devices Dumps Library, by providing more information.

A daemon to control NUT features locally or remotely:

• scan,
• configuration,
• daemons start / stop / restart / reload.

Use cases:

• mass configuration (including initial configuration) of a network of NUT systems,
• automated initial configuration of system with USB devices.

@balooloo: to be completed

APC BackUPS LS 500 (USB HID PDC) has a battery-not-present state that is not mapped to ups.status (as of v.2.7.1)

Create a generic driver to support Modbus devices.

It will:

• use libmodbus (http://libmodbus.org)
• support both RTU (RS232/485) and TCP communication
• support various devices (ups, meters, ...)
• use external definition files, including default communication settings (for RTU), and data mapping

A preliminary support has already been committed:
https://github.com/networkupstools/nut/tree/nutdrv-modbus

FIXME: Need completion from Arno

libnutconf and nutconf are available in branch libconf:
https://github.com/networkupstools/nut/tree/libconf

Error messages are generated at LOG_NOTICE when PID files can't be opened. In some cases (checking for existing processes, etc.) these "errors" are expected. Also, upsmon apparently does not remove its PID file when it exits, but upsd does.

http://thread.gmane.org/gmane.comp.monitoring.nut.user/8019

Log the error code here:

https://github.com/networkupstools/nut/blob/v2.7.1/clients/upsmon.c#L1683

The Linux system(2) man page doesn't say that errno is set, but we should log the return code if system() fails.

(Theoretically, this applies to NetBSD and OpenBSD as well, but FreeBSD's USB stack seems more complete at the moment.)

Some code: https://github.com/clepple/nut/tree/devd-usb

Things to do:

• Generate devd.conf files for known USB devices
• Install devd.conf files in the right place (includes detection of *BSD)
• Document the installation process (restarting devd, other permissions issues like scanning the bus, etc.)
• Coordinate with ports maintainer

FAQ needs an overall refresh.
Some important topics may still be missing (see commit eb1f01e) or not enough visible, others may not be applicable anymore, ...

A few things are still missing, to complete IPMI support:

• IPMI scan should first do an "ipmi ping", prior to doing an actual ipmi open.
  Refer to http://www.gnu.org/software/freeipmi/manpages/man8/ipmiping.8.html
  This is applicable to both nut-ipmipsu and nut-scanner
• IPMI driver still does not handle commands (on, off, sw off)
  Refer to http://www.gnu.org/software/freeipmi/manpages/man8/ipmipower.8.html

Lower priority:

• better NUT IPMI back-ends interface (create a NUT IPMI API)
• Allow runtime selection (beside compile time) of the IPMI backend used
• 2nd IPMI back-end implementation, beside of FreeIPMI (ipmiutil or ipmitool)
• Rename to nutdrv-ipmipsu (or simply nutdrv-ipmi?)

Note on ipmiutil:
It provides transparent IPMI command-level access for in-band and out-of-band.
Using the ipmi_sample.c in the devel package does support FRU, SDR, etc., so that shows how this is used, and could be compiled as a library as well. And there is an ipmi_sample_evt.c for handling events.
Andy Cress also proposed some help and is interested in feedback.

Note on ipmitool:
Used at least on Opengear systems.

When building nut 2.7.1 on debian, I get the following warnings:

dpkg-shlibdeps: warning: symbol upsdebugx used by debian/libupsclient3/lib/x86_64-linux-gnu/libupsclient.so.3.0.1 found in none of the libraries
dpkg-shlibdeps: warning: symbol upslogx used by debian/libupsclient3/lib/x86_64-linux-gnu/libupsclient.so.3.0.1 found in none of the libraries
dpkg-shlibdeps: warning: symbol xmalloc used by debian/libupsclient3/lib/x86_64-linux-gnu/libupsclient.so.3.0.1 found in none of the libraries
dpkg-shlibdeps: warning: symbol xstrdup used by debian/libupsclient3/lib/x86_64-linux-gnu/libupsclient.so.3.0.1 found in none of the libraries

I guess that libupsclient should link against libcommon

This stems from not including nut_version.h in the distributed tarball. However, it was removed due to this Debian bug which I don't completely understand:

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=613643

Further discussion on nut-upsdev:

http://article.gmane.org/gmane.comp.monitoring.nut.devel/6388

Several drivers have some boilerplate code that implements a boolean type through an enum or an int. Some, such as the one in mge-utalk.h, are slightly different from the rest. This should be moved to a common header file.

Complete port of NUT to MS Windows.

Alioth feature request:
https://alioth.debian.org/tracker/index.php?func=detail&aid=302177&group_id=30602&atid=411545

Tripp Lite has provided a list of models and Protocol numbers:

http://article.gmane.org/gmane.comp.monitoring.nut.user/8173

The Protocol number for HID PDC devices is equal to the USB Product ID, so this should be reflected in the detection routine and documentation.

2nd generation of lbnutscanner and nut-scanner:

• plugin design:
  • each "bus" is provided through a specific library (aka "plugin"),
  • each plugin provides everything: help, init, scan, cleanup/close,
  • plugins using weak dependencies through dl_* functions (no hard link)
  • ip address iterator is also perfectible
• libnutscanner provides .so listing and loading (to wrap available and usable plugins)
• nut-scanner is almost empty shell

This will solve the #ifdef issue, and make nut scan library and tool more maintainable and usable!

NUT Devices Dumps Library (http://networkupstools.org/ddl/) needs some more actions before being complete:

• Reference on the website Compatibility page (@aquette)
• Reference on the website Documentation page (@aquette)
• Create a dedicated repository (https://github.com/networkupstools/nut-ddl) and website (http://networkupstools.org/ddl/)
• Create a better page, or set of pages, to document / search / submit dumps ( @zykh, would you be able to help?)
• Harvest the web and NUT mailing lists for existing devices dumps (@zykh addressed that)

Postponed items

• Complete task #94 (NUT Dummy driver Gen3)
• Call to users for massive report (who?) => or make it easier to submit DDL report.
  Part of this was started to be addressed with ca9f601

Some of the UPS model and cable names turn into telephone numbers on Safari on iPhone, especially on the HCL.

Add something like this:

<meta name="format-detection" content="telephone=no" />

• in usbhid-ups spirit (2nd generation of meta driver, with sub-drivers)
• support external MIB2NUT files (externalize all mib2nut?)
  Already available through the DMF
• support SNMP traps ("events notification")
  • how to handle traps for multiple drivers?
  • ==> nutdrv-trap{d,handler,dispatcher,whatever_name} is a driver that receives traps and dispatches it over the nutdrv-snmp driver(s) instance(s), using ups.conf->port field as a key. The main issue here is that this driver needs to be started (not forcibly declared) somewhere. nutdrv-controller could do the job, with a specific hook
  • ==> nutd could do the job of starting. but that would not be clean!

using ESR' reposurgeon

Hello up there, NUT people.

Here is AL175 driver patch, updated again, based on feedback from 2009:

http://lists.alioth.debian.org/pipermail/nut-upsdev/2008-December/003641.html
http://lists.alioth.debian.org/pipermail/nut-upsdev/2008-December/003644.html
http://lists.alioth.debian.org/pipermail/nut-upsdev/2009-January/003710.html
http://lists.alioth.debian.org/pipermail/nut-upsdev/2009-January/003778.html
http://lists.alioth.debian.org/pipermail/nut-upsdev/2009-January/003779.html
http://lists.alioth.debian.org/pipermail/nut-upsdev/2009-February/003791.html
http://lists.alioth.debian.org/pipermail/nut-upsdev/2009-February/003790.html

(lists.a.d.o is currently down, but the links should be ok)

I've tried to clean it up, but unfortunately no longer have hardware
to test and will not have any in foreseeable future, so the driver was reworked
to meet the project code quality criteria, without testing on real hardware.
Some bugs may have crept in.

The patches are here:

http://repo.or.cz/w/networkupstools/kirr.git/commitdiff/dd43abcf0292a99b93346522c364f417616b58df
http://repo.or.cz/w/networkupstools/kirr.git/commitdiff/f2ab786359179970d484b447738b1cecc8461e05

available for pull from repo.or.cz repository as

git pull git://repo.or.cz/networkupstools/kirr.git y/al175-2013

If you would ask me, there is no reason not to merge it - yes, the code could be seen as somewhat unusual, but was (when testing with hardware) functional and is structured and logic and flow could be traced. Several people already asked me about al175 driver on email, so I think rejecting the driver benefits noone.

Thanks beforehand,
Kirill

Hi,

With the new 2.7.1 version, the POWERDOWNFLAG path has changed from /etc/killpower to @CONFPATH@/killpower

This means that the new default path is /etc/nut/killpower on most of the distributions.

I'm wondering if this shouldn't be reverted to /etc/killpower

While the selectors for the jQuery filters on the HCL do not need to be displayed if JavaScript is disabled, the key which shows the meaning of the colors should be enabled.

Similarly, we should check to see that this makes sense in something other than Mozilla-based or Webkit-based browsers.

Do you will support http://www.fsp-europe.com/ep850.php or series FORTRON FSP "Line Interactive"?

upsc, upscmd and upsrw should provide a -t timeout option.
Underlying mechanism is already available in libupsclient->upscli_tryconnect(...).

Alioth patch #313881
https://alioth.debian.org/tracker/index.php?func=detail&aid=313881&group_id=30602&atid=411544

An external contributor (Imene Guerfa) is already working on this.

Alioth bug #313884
http://lists.alioth.debian.org/pipermail/nut-upsuser/2012-October/007994.html

While the target apcupsd host is down, there are spurious stale/not-stale messages in syslog:

09:10:21 hostname apcupsd-ups[94147]: can't communicate with apcupsd!
09:10:21 hostname upsd[94149]: UPS [apcupsd] data is no longer stale
09:10:37 hostname upsd[94149]: Data for UPS [apcupsd] is stale - check driver

Verbose is good, but in this case, we should probably make sure the staleness is being handled properly in other cases, and when the host comes back up.

Manufacturer: MicroDowell
Model: B.Box LP 500

ups.conf

[microdowell]
        driver = genericups
        port = /dev/ttyS0
        upstype = 7

upsc

$ upsc microdowell
device.mfr: CyberPower
device.model: Power99
device.type: ups
driver.name: genericups
driver.parameter.pollinterval: 2
driver.parameter.port: /dev/ttyS0
driver.parameter.upstype: 7
driver.version: 2.6.3
driver.version.internal: 1.36
ups.mfr: CyberPower
ups.model: Power99
ups.status: OL

OL, OB and LB is working with upstype 7 [CP=RTS] [OL=CTS] [LB=-DCD].

UPS shutdown with upstype 7 [SD=DTR]

$ sudo upsdrvctl shutdown microdowell
Network UPS Tools - UPS driver controller 2.6.3
Network UPS Tools - Generic contact-closure UPS driver 1.36 (2.6.3)
UPS type: CyberPower Power99

Initiating UPS shutdown

UPS shutdown only works when on-battery and has a delay of about 1min until execution (something between 50sec to 1min 30sec on mine).

References:
http://www.ezdirect.it/pdf/lp500.pdf
https://forums.gentoo.org/viewtopic-t-730172-start-0.html

Edit:
I forgot to mention that I also tried the windows software UPSEye from MicroDowell, it also can't do more than displaying the UPS state without detail information. Therefore it seems that genericups is the best protocol the UPS supports. Moreover the datasheet mentions "RS232 Simple Contacts interfaces".

Integrate the QRT work done for Ubuntu in NUT mainline.
Quoting docs/nut-qa.txt:

a runtime testing suite, which automates the inter layer communication testing
(driver - upsd - upsmon / clients), that is part of Ubuntu.
The NUT testing script is available in the Ubuntu QA Regression Testing suite:

• http://bazaar.launchpad.net/~ubuntu-bugcontrol/qa-regression-testing/master/view/head:/scripts/test-nut.py => https://git.launchpad.net/ubuntu/+source/nut/tree/debian/tests
• https://code.edge.launchpad.net/qa-regression-testing.

It installs nut, configure with for the dummy-ups driver, changes a few data and
check that these are well propagated with upsc.

This probably involves testlib.py, but should be limited if it goes beyond!

NOTE: it's a good time to reconsider QRT as a whole!
Other framework may now be more suitable to test our broader scope (OS at least)

Also hook it to the "check" target, for buildbot use.

Versions prior to NSS merge provided the following output:

nut-2.6.4$ ./clients/upsc -L
test: Description unavailable

Since the merge of NSS, we now have (here using the latest git master):

$ ./clients/upsc -L
upscli not initialized, force initialisation without SSL configuration
Init SSL without certificate database
Can not connect to localhost in SSL, continue uncrypted
test: Description unavailable

This should be fixed prior to the next release, since it breaks compatibility with, for example, scripts and tools using upsc output.

The issue lies in clients/upsclient.c, when tryssl (default) is set and SSL can't be enabled...

USB support on Solaris exists for years.
But it has never been fully completed.

Some information:

• base README info: https://gist.github.com/aquette/5068233
• Alioth patch #310561:
  https://alioth.debian.org/tracker/index.php?func=detail&aid=310561&group_id=30602&atid=411544

Things to do:

• Generate a script that executes add_drv/update_drv/rem_drv (step 4-c of the above gist) for known USB devices (used for Solaris package postint/prerm)
• Integrate the script (and other actions) in Solaris package
• Integrate the remaining documentation info, if any

There are many patches out there, and very probably some interesting:
https://cvsweb.openbsd.org/ports/sysutils/nut/patches/ replicated at https://github.com/openbsd/ports/tree/master/sysutils/nut

(Previously http://www.openbsd.org/cgi-bin/cvsweb/ports/sysutils/nut/patches/ was the place)

Compiling usbhid-ups.c fails.

This was first reported for OpenBSD 4.8: http://thread.gmane.org/gmane.comp.monitoring.nut.user/6073

And still in 5.3: http://thread.gmane.org/gmane.comp.monitoring.nut.user/8006/focus=8017

It is probably just a matter of commenting out that case, but someone with OpenBSD should check to see what other error codes are used for USB errors.

Some of the warnings on the clang builders prompted me to take a closer look at the libconf branch.

• https://github.com/networkupstools/nut/blob/libconf/common/nutstream.cpp#L346

NutStream::status_t NutFile::putData(const std::string & data) has a comment that there is not a FILE* interface for writing arbitrary data. What about fwrite()?

• https://github.com/networkupstools/nut/blob/libconf/common/nutstream.cpp#L211

tempnam() should probably be replaced with mkstemp(). Then you won't need the hard-coded /var/tmp variable. (Some systems implement per-user $TMPDIR variables, which eliminate some corner cases and race conditions.)

• https://github.com/networkupstools/nut/blob/libconf/common/nutstream.cpp#L488

This expression won't do bit operations: ss << (packed && 0x000000ff) << ".";

Also, you should consider using the standard inet_?to? functions here, especially for the IPv6 address printer.

When parsing the overrides for OB/LB/CP/SD, show names as they are encountered. Unknown names just fall through, so a typo in a signal name might go unnoticed.

I've upgraded from 2.6.4 to 2.7.1 and in the upgrade on some servers I lost connectivity to the Evolution UPS serial port. I tracked this down to the patch 65db105, reverting it made things work again.