Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.debian.user > #205023 > unrolled thread

shell script problem

Back to article view | Back to linux.debian.user

Contents

  shell script problem ghe <ghe@slsware.net> - 2019-02-05 16:50 +0100
    Re: shell script problem john doe <johndoe65534@mail.com> - 2019-02-05 17:20 +0100
      Re: shell script problem ghe <ghe@slsware.net> - 2019-02-05 20:00 +0100
        Re: shell script problem Greg Wooledge <wooledg@eeg.ccf.org> - 2019-02-05 20:10 +0100
          Re: shell script problem ghe <ghe@slsware.net> - 2019-02-05 22:10 +0100
            Re: shell script problem Greg Wooledge <wooledg@eeg.ccf.org> - 2019-02-05 22:10 +0100
              Re: shell script problem ghe <ghe@slsware.net> - 2019-02-05 22:30 +0100
    Re: shell script problem ghe <ghe@slsware.net> - 2019-02-05 19:20 +0100
      Re: shell script problem john doe <johndoe65534@mail.com> - 2019-02-05 19:30 +0100
      Re: shell script problem Jude DaShiell <jdashiel@panix.com> - 2019-02-05 20:30 +0100
      Re: shell script problem John Crawley <john@bunsenlabs.org> - 2019-02-06 06:30 +0100
        Re: shell script problem <tomas@tuxteam.de> - 2019-02-06 11:20 +0100
        Re: shell script problem Rusi Mody <rustompmody@gmail.com> - 2019-02-06 11:20 +0100
        Re: shell script problem Greg Wooledge <wooledg@eeg.ccf.org> - 2019-02-06 15:00 +0100

#205023 — shell script problem

Buster

A shell script has begun to throw errors I can't seem to get rid of.
Didn't used to -- it started in the past few days.

I have access to a couple Internet connections: a slow but extremely
reliable T1 I've been with for years, and a Comcast residential cable
WiFi that's odd, but very fast. There are 2 Ethernet ports on the box.

I wrote a shell script for the frequent and large Buster updates, etc.
that brings up the Comcast WiFi, makes changes in the DHCP info,
modifies the routing table so the mirrors in the sources list are routed
through Comcast, and fires off aptitude.

The problem is that toward the end of bring up Comcast, I get a message
from systemd saying '-.mount is masked'. The systemd man pages say there
is indeed something called -.mount and that being masked will kill any
process encountering a masked something. The message says it's going to
quit the script.

Then it goes on with the script, and as best I can tell, finishes properly.

The man pages say there is a way to unmask things (I can't remember what
it is). When I run the command, I'm told that the parameter is wrong (it
doesn't like the '-', understandable). I put it in quotes, same thing. I
try it with something else masked in the list, and it comes back with
the expected *nix response when all is well (nothing). But when I look
at the list again, it's still masked.

What's going on? What does 'masked' mean? What is '-.mount'? Should I
just ignore the error message? Can I ignore any systemd error message?

-- 
Glenn English

[toc] | [next] | [standalone]

#205025

On 2/5/2019 4:43 PM, ghe wrote:
> Buster
>
> A shell script has begun to throw errors I can't seem to get rid of.
> Didn't used to -- it started in the past few days.
>
> I have access to a couple Internet connections: a slow but extremely
> reliable T1 I've been with for years, and a Comcast residential cable
> WiFi that's odd, but very fast. There are 2 Ethernet ports on the box.
>
> I wrote a shell script for the frequent and large Buster updates, etc.
> that brings up the Comcast WiFi, makes changes in the DHCP info,
> modifies the routing table so the mirrors in the sources list are routed
> through Comcast, and fires off aptitude.
>
> The problem is that toward the end of bring up Comcast, I get a message
> from systemd saying '-.mount is masked'. The systemd man pages say there
> is indeed something called -.mount and that being masked will kill any
> process encountering a masked something. The message says it's going to
> quit the script.
>

What message?

> Then it goes on with the script, and as best I can tell, finishes properly.
>
> The man pages say there is a way to unmask things (I can't remember what

$systemctl unmask <SERVICE-NAME>

> it is). When I run the command, I'm told that the parameter is wrong (it

Which cmd?

> doesn't like the '-', understandable). I put it in quotes, same thing. I
> try it with something else masked in the list, and it comes back with
> the expected *nix response when all is well (nothing). But when I look
> at the list again, it's still masked.
>
> What's going on? What does 'masked' mean? What is '-.mount'? Should I
> just ignore the error message? Can I ignore any systemd error message?
>

- What's going on -- can't say without seeing the script in question and
the error message you're getting.


In general, ignoring error messages is a recipe for disaster.

--
John Doe

[toc] | [prev] | [next] | [standalone]

#205032

On 2/5/19 9:12 AM, john doe wrote:

> What message
Never mind. It quit complaining. No changes to the scripts. Maybe a
systemd update -- there are several updates to lots of code with testing.

Thanks for the responses.

> In general, ignoring error messages is a recipe for disaster.

You'd think so. Usually, though, they aren't just kidding...

I'd still like to know what 'mask' means in systemd. And what a
'-.mount' is.

-- 
Glenn English

[toc] | [prev] | [next] | [standalone]

#205034

On Tue, Feb 05, 2019 at 11:51:42AM -0700, ghe wrote:
> I'd still like to know what 'mask' means in systemd.

>From man systemctl:

       mask NAME...
           Mask one or more units, as specified on the command line. This will
           link these unit files to /dev/null, making it impossible to start
           them. This is a stronger version of disable, since it prohibits all
           kinds of activation of the unit, including enablement and manual
           activation. Use this option with care. This honors the --runtime
           option to only mask temporarily until the next reboot of the
           system. The --now option may be used to ensure that the units are
           also stopped. This command expects valid unit names only, it does
           not accept unit file paths.

[toc] | [prev] | [next] | [standalone]

#205040

On 2/5/19 12:01 PM, Greg Wooledge wrote:

>>From man systemctl:
> 
>        mask NAME...
>            Mask one or more units, as specified on the command line. This will
>            link these unit files to /dev/null, making it impossible to start
>            them. This is a stronger version of disable, since it prohibits all
>            kinds of activation of the unit, including enablement and manual
>            activation. Use this option with care. This honors the --runtime
>            option to only mask temporarily until the next reboot of the
>            system. The --now option may be used to ensure that the units are
>            also stopped. This command expects valid unit names only, it does
>            not accept unit file paths.

Yeah, I saw that. So mask is stronger that whatever disable is. And it
'prohibits all kinds of activation of the unit' -- but it, apparently,
encountered a masked unit, notified me, them just went on with the
script. I've seen error messages saying '<something is bent> but going
on anyway'. Nothing like that in this case, but it did go on anyway.
Apparently.

This adventure has made me believe that systemd was a massive effort.
But also that there are some areas where it could still use a little work.

I still think that that massive effort might have been better spent
fixing init. Handling threads, tightening some methods and regulations,
tidying up init.d and its buds, enabling startups in PERL, Python, Ruby,
C, Java, etc. -- things legible and debuggable. Maybe even shell, if
you're into a syntax from the late 1940s...

-- 
Glenn English

[toc] | [prev] | [next] | [standalone]

#205041

On Tue, Feb 05, 2019 at 02:00:12PM -0700, ghe wrote:
> I still think that that massive effort might have been better spent
> fixing init. Handling threads, tightening some methods and regulations,
> tidying up init.d and its buds, [...]

Start here: https://jdebp.eu/FGA/system-5-rc-problems.html

[toc] | [prev] | [next] | [standalone]

#205042

On 2/5/19 2:05 PM, Greg Wooledge wrote:

>> I still think that that massive effort might have been better spent
>> fixing init. Handling threads, tightening some methods and regulations,
>> tidying up init.d and its buds, [...]
> 
> Start here: https://jdebp.eu/FGA/system-5-rc-problems.html

Yup. The init startups and such were, indeed, at best bent. So fix them.
There's something to be said for sticking with Thompson and Ritchie's
thoughts when developing *nix. They were smart, thoughtful guys.

-- 
Glenn English

[toc] | [prev] | [next] | [standalone]

#205030

On 2/5/19 9:19 AM, Jude DaShiell wrote:

> Have you tried replacing "-" with \45 yet?  That's the ascii equivalent
> for "-'.

Excellent idea. But:

root@sbox:~# systemctl unmask \45.mount
Unit 45.mount does not exist, proceeding anyway.

(Same with quotes.)

-- 
Glenn English

[toc] | [prev] | [next] | [standalone]

#205031

On 2/5/2019 7:17 PM, ghe wrote:
> On 2/5/19 9:19 AM, Jude DaShiell wrote:
>
>> Have you tried replacing "-" with \45 yet?  That's the ascii equivalent
>> for "-'.
>
> Excellent idea. But:
>
> root@sbox:~# systemctl unmask \45.mount
> Unit 45.mount does not exist, proceeding anyway.
>
> (Same with quotes.)
>

$ systemctl status -- -.mount

--
John Doe

[toc] | [prev] | [next] | [standalone]

#205035

Okay so "-" isn't a systemd unit name.  At this point, you might do well
to examine dmesg.  Check for fatals errors and warnings and see if any
interesting material turns up.

On Tue, 5 Feb 2019, ghe wrote:

> Date: Tue, 5 Feb 2019 13:17:49
> From: ghe <ghe@slsware.net>
> To: debian-user@lists.debian.org
> Subject: Re: shell script problem
> Resent-Date: Tue,  5 Feb 2019 18:18:34 +0000 (UTC)
> Resent-From: debian-user@lists.debian.org
>
> On 2/5/19 9:19 AM, Jude DaShiell wrote:
>
> > Have you tried replacing "-" with \45 yet?  That's the ascii equivalent
> > for "-'.
>
> Excellent idea. But:
>
> root@sbox:~# systemctl unmask \45.mount
> Unit 45.mount does not exist, proceeding anyway.
>
> (Same with quotes.)
>
>

-- 

[toc] | [prev] | [next] | [standalone]

#205051

On 06/02/2019 03.17, ghe wrote:
> On 2/5/19 9:19 AM, Jude DaShiell wrote:
> 
>> Have you tried replacing "-" with \45 yet?  That's the ascii equivalent
>> for "-'.
> 
> Excellent idea. But:
> 
> root@sbox:~# systemctl unmask \45.mount
> Unit 45.mount does not exist, proceeding anyway.
> 
> (Same with quotes.)
> 
You might try
root@sbox:~# systemctl unmask $'\45'.mount
Just a thought.
-- 
John

[toc] | [prev] | [next] | [standalone]

#205055

[Multipart message — attachments visible in raw view] — view raw

On Wed, Feb 06, 2019 at 02:04:54AM -0800, Rusi Mody wrote:
> I seem to be missing something... 
> Is -.mount literally a thing? 
> Or -- more likely -- are shell's arcane quoting/interpolating rules garbling something quite into a "-" unit? 

Perhaps you're better off looking into the systemd
config directories and finding out what symlinks point
to /dev/null or something (no systemd expert here, though).

Cheers
-- t

[toc] | [prev] | [next] | [standalone]

#205056

I seem to be missing something... 
Is -.mount literally a thing? 
Or -- more likely -- are shell's arcane quoting/interpolating rules garbling something quite into a "-" unit? 

[toc] | [prev] | [next] | [standalone]

#205061

On Wed, Feb 06, 2019 at 02:22:35PM +0900, John Crawley wrote:
> On 06/02/2019 03.17, ghe wrote:
> > On 2/5/19 9:19 AM, Jude DaShiell wrote:
> > > Have you tried replacing "-" with \45 yet?  That's the ascii equivalent
> > > for "-'.
> > 
> > Excellent idea. But:
> > 
> > root@sbox:~# systemctl unmask \45.mount
> > Unit 45.mount does not exist, proceeding anyway.
> > 
> > (Same with quotes.)
> > 
> You might try
> root@sbox:~# systemctl unmask $'\45'.mount
> Just a thought.

You're all just making stuff up without understanding it.

First of all, the ASCII value of - is 45 decimal, or 055 octal, or 0x2d
hexadecimal.  If you use bash's $'\...' quoting syntax, you get to choose
between octal and hexadecimal, the default being octal.

So, $'\055' is equivalent to '-'.  $'\45' is a percent sign.

Second, whatever form of quoting you use in your shell command is simply
a mechanism to make sure your intended arguments get passed to the
command you're invoking.  Once the shell has parsed your quoting, the
quotes are removed.  The command doesn't see them.

As arguments, - and "-" and '-' and $'-' and $'\055' and $'\x2d' are all
exactly the same.  There is no difference at all.

wooledg:~$ args - "-" '-' $'-' $'\055' $'\x2d'
6 args: <-> <-> <-> <-> <-> <->

The command (mount, umount, systemctl unmask, or whatever) does not know
or care which kind of quoting, if any, you used in your shell.

This is also why the classic Unix shell FAQ "how do I remove a file whose
name starts with -" is such a puzzler for many people.  People think
that by quoting the leading - they will somehow make rm treat it as a
filename instead of an option.  But because of how shells *work*, the
quotes are removed, and rm has no idea that you quoted it.

Answer to the FAQ: you must tell rm that this argument is a filename
and not an option.  You do that by putting the end-of-options indicator --
before the filename.  E.g. rm -- -myfile

Another way to do it is to give rm a pathname that does not begin with 
a - character.  E.g. rm ./-myfile

The latter only works because the argument to rm *is* a pathname,
and relative pathnames can have a leading ./ prepended to them without
changing their meaning.

Now, I don't know how systemctl unmask works.  You can try putting --
before your unit argument, if the unit name truly begins with a -,
and see if that helps.  Or, you can try to figure out whether this is
really the correct unit name at all.  It seems rather dubious.

Is there any chance that your fstab file is simply malformed, and is
being converted into bogus .mount units?  Maybe you should go through
your /etc/fstab and look for errors.

[toc] | [prev] | [standalone]

Back to top | Article view | linux.debian.user

csiph-web