gpsd-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [gpsd-users] gpsd systemd troubleshooting: Draft


From: Gary E. Miller
Subject: Re: [gpsd-users] gpsd systemd troubleshooting: Draft
Date: Fri, 4 Oct 2019 18:39:26 -0700

Yo Charles!

On Fri, 4 Oct 2019 19:07:09 -0600
Charles Curley <address@hidden> wrote:

> On Fri, 4 Oct 2019 16:02:41 -0700
> "Gary E. Miller" <address@hidden> wrote:
> 
> > The order of the new material is loopy.  First it talks about the
> > config files that systemd(ung) uses, then talks about installing
> > gpsd with apt, which would install the files that the first part
> > expected to be present.  
> 
> The order of the material is 1) introduction, which is the more
> general stuff; and 2) a specific example. That is, from the general
> to the specific, which is how documentation usually proceeds.

Yes, which is what I complained about.  This is a HOWTO, not an
overview.

> > And, no need to talk about the installation in the troubleshooting
> > doc. The installation details go in the INSTALL file, just point to
> > them.
> >
> > The doc also assumes a user has installed with "apt", which is not
> > present on many systemd(ick) systems.  The troubleshooting needs to
> > support whatever method, from the INSTALL file, that the system is
> > in.  Possibly a mess of distro and self-installed gpsd.  
> 
> 
> OK. Something like, "assuming you have already installed gpsd and the
> client xgps or cgps....". I do think I should call out which
> components I expect to the reader to have handy early in the example.

Never assume.  Tell the reader he must have completed the steps in the
distro specific section of the INSTALL file.

Then start with a few tests (gpsd -V) so the user can confirm that
was done properly.  Amazing how often just that fails, and they are
complaining about ntpmon!

> Debian packaging is common enough in the Linux world that most users
> can translate an "apt install" to their preferred package manager. But
> no matter, we'll leave that to the reader.

Uh, no.  I never ever touch apt.  I have no idea how it works, nor
do most users.   It is distro specific enough that it belongs in the
distro specific INSTALL section, not in the troubleshooting doc.

Just take a look at the INSTALL file for how many different ways
there are to install distro packages.

Start with simple tests that confirm a standard, or local, install of
gpsd and that the required options are compiled in.

Then proceed to show how to use/debug systemd(arth).

> I used a Debian example because that's what I had in front of me.

yes, write what you know.  By systemd, systemctl, the service files, etc.
are not Debian specific, they are part of the systemd package.  Just like
xgps is not Debian specific, unless they broke it.

It might be a choice to put the generic systemd(astdardly) stuff in the
INSTALL file, as configuring services files, etc. is part of installing.
How that partitions is much less obvious...

> I
> could instead write up installing on a Debian box from the git repo,
> but will have to actually do that first. That may take a few days.

No need.  That is the point of the INSTALL file, you are working on the
troubelshooting doc.  Add just enough checks, at the top, to verify
a good installation, then move on to the systemd(reck) stuff.

> > The doc assumes the user is usuing a hot-pluggable GNSS receiver.
> > The majority case case of users with problems is Raspberry Pi users
> > that do not have hot-pluggable receivers.  
> 
> Ah, now that may be a problem. I found out that to get the changes to
> the gpsd.socket file to take effect, it was not enough to run
> "systemctl stop ... ; systemctl start ...". I had to unplug the
> receiver.

Yes, that would be a problem.  Maybe someone that suffers systemd(umbest)
can shed some light on that.


> I found that annoying and filed it away for future
> examination.

One step at a time.

> As your discussion of the Raspberry Pi makes clear, I
> should not have had to unplug it. This may be another systemd related
> issue.

Dunno, hopefully someone that uses systemd can provide some insight.

> One workaround may be to kill gpsd between the "stop" and the "start".
> I'll experiment with that.

Let us know what you find.

> If most of the issues are Raspberry Pi related, maybe the example
> should be on one? Or is Debian close enough?

Uh, Debian is a distro, Raspberry Pi is a platform.  Different things.

The systemd(ung) stuff should be similar enough on both platforms.

Luckily my Raspberry Pi's are systemd(isaster)-less.

> Any Raspberry Pi users care to chime in?

I think most people use Raspbian on the RasPi, so any Debian specifics
should be close.  Those of us not using systemd(ank) are blissfully
unaffected!

RGDS
GARY
---------------------------------------------------------------------------
Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97703
        address@hidden  Tel:+1 541 382 8588

            Veritas liberabit vos. -- Quid est veritas?
    "If you can't measure it, you can't improve it." - Lord Kelvin

Attachment: pgptqudoObmIH.pgp
Description: OpenPGP digital signature


reply via email to

[Prev in Thread] Current Thread [Next in Thread]