Re: bewildering boot problem -- a little more info

[tomas]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

On Mon, Oct 05, 2015 at 11:05:20PM -0600, Glenn English wrote:

[...]

> > Hmmm. Sorry to be so unspecific.
> 
> No prob. Everybody's happy now. I just have some data to copy over from the 
> old disk. At least it wasn't the hardware...

In any case, glad you solved it :-)

regards
- -- t
-BEGIN PGP SIGNATURE-
Version: GnuPG v1.4.12 (GNU/Linux)

iEYEARECAAYFAlYTaO4ACgkQBcgs9XrR2kZ0uQCffpzX+e3a0RKH3LYbRYxR4PfA
E4UAnjT4Y1tWNs9wDqf8bC6SpjVmnuid
=Vkux
-END PGP SIGNATURE-

Re: bewildering boot problem -- a little more info

[tomas]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

On Mon, Oct 05, 2015 at 11:53:06AM -0600, Glenn English wrote:
> 
> On Oct 5, 2015, at 1:04 AM, to...@tuxteam.de wrote:
> 
> > Then try "ssh @localhost".
> 
> Works, kinda. It does the same thing to localhost or the hostname, from the 
> console or from a terminal in XFCE. (I've got SSH running with keys around 
> the local net(s) (no login required)). The MOTD comes up right away, but 
> there's a long delay before the shell prompt comes up. 

Hmm. Weird. I dimly remember that tcpwrappers did something
similar: to check the host name they sometimes tried a
reverse host lookup, which took its time when it failed.

But perhaps I'm up the wrong alley.

Possibly watching the connection attempt with tcpdump or
wireshark sheds light on this.

The behavour of ssh you describe is weird indeed -- the
MOTD must be coming from the server side, so the connection
is established already. As if the shell were checking something
before it decides to talk to you.

Hmmm. Sorry to be so unspecific.

regards
- -- tomás
-BEGIN PGP SIGNATURE-
Version: GnuPG v1.4.12 (GNU/Linux)

iEYEARECAAYFAlYSuRkACgkQBcgs9XrR2kanRACcDe2WUz17WUUBnX4nVmO3r+NM
ly4AnRXwUaIlYD5RqSw+lsflkqArgEf6
=WQFH
-END PGP SIGNATURE-

Re: bewildering boot problem

[rlharris]

On Mon, October 5, 2015 12:53 pm, Glenn English wrote:
> It seems to me that there's a networking problem, but
> intermittent and from several directions. I don't understand it at all.
>
> Maybe it's hardware...

Have you had a lightning storm in the area recently?  And is everything
connected to the machine plugged into a power strip with a surge arrestor?

Reportedly, most lightning damage to computer gear is due to transients
entering through the telephone line, rather than from the power line.

And even without nearby lightning strokes, transients are found on power
lines, due to connection and disconnection of inductive and capacitive
loads.

Electrical transients can cause cumulative damage without causing
immediate failure.  A good analogy is the damage inflicted upon a large
boulder by repeatedly striking it with a sledge.  Even after many blows,
no damage may be apparent, but eventually a blow shatters the boulder,
because the damage is cumulative.  In this respect, solid-state devices
are like the boulder.

By the way, a surge arrestor also suffers cumulative damage; but it is
meant to be sacrificial.

Russ

Re: bewildering boot problem -- a little more info

[Glenn English]

On Oct 5, 2015, at 1:04 AM, to...@tuxteam.de wrote:

> Then try "ssh @localhost".

Works, kinda. It does the same thing to localhost or the hostname, from the 
console or from a terminal in XFCE. (I've got SSH running with keys around the 
local net(s) (no login required)). The MOTD comes up right away, but there's a 
long delay before the shell prompt comes up. 

startx takes a very long time and prints some error messages, but eventually 
runs XFCE as expected. Minicom works from a terminal in XFCE, but not from the 
console. iwconfig says wlan0 is connected to my wifi AP (Ethernet isn't 
available where the laptop is). A ping to a host on the LAN says there's no 
route available -- but there are a couple valid ones in the table.

OTOH, the kernel's startup messages are gone this morning. I'm thinking a bare 
metal reinstall is in order. One or more things are really bent in there 
somewhere. It seems to me that there's a networking problem, but intermittent 
and from several directions. I don't understand it at all.

Maybe it's hardware...

Thanks, everyone, for the suggestions.

-- 
Glenn English

Re: bewildering boot problem -- a little more info

[Felix Miata]

Glenn English composed on 2015-10-04 18:14 (UTC-0600):

> I disabled XDM (sure is nice to have text config files) and rebooted. It
> came up at a regular login prompt, and no error messages. startx took a
> *very* long time -- said it couldn't find the hostname (something like
> that) -- but it eventually started XFCE. And the messages were back.

> Something I forgot to mention: When XFCE starts and I open a terminal, the
> terminal comes up right away, but it takes  a long time (30+ seconds, it
> feels like) to display the prompt.

Do you have an entry in /etc/hosts matching your actual IP to the content in
/etc/hostname when you observe this delay?
-- 
"The wise are known for their understanding, and pleasant
words are persuasive." Proverbs 16:21 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409 ** a11y rocks!

Felix Miata  ***  http://fm.no-ip.com/

Re: bewildering boot problem -- a little more info

[Glenn English]

On Oct 5, 2015, at 11:53 AM, to...@tuxteam.de wrote:

> Hmm. Weird. I dimly remember that tcpwrappers did something
> similar: to check the host name they sometimes tried a
> reverse host lookup, which took its time when it failed.

I've never seen or heard of anything like this -- the kernel complaining of the 
wrong #defines in the source code? That, to me, is massively broken. Anyway, 
it's all OK now; it got a new Jessie this afternoon :-)

> But perhaps I'm up the wrong alley.

Probably. I claim there were just too many wrong alleys. And certainly not 
worth the time to try to fix it.

> Hmmm. Sorry to be so unspecific.

No prob. Everybody's happy now. I just have some data to copy over from the old 
disk. At least it wasn't the hardware...

-- 
Glenn English

Re: bewildering boot problem -- a little more info

[Glenn English]

On Oct 5, 2015, at 12:51 PM, Felix Miata  wrote:

> Do you have an entry in /etc/hosts matching your actual IP to the content in
> /etc/hostname when you observe this delay?

Yup. I rely on the host files, so I keep then correct and accurate.

-- 
Glenn English

Re: bewildering boot problem

[Glenn English]

On Oct 5, 2015, at 12:45 PM, rlhar...@oplink.net wrote:

> On Mon, October 5, 2015 12:53 pm, Glenn English wrote:

>> Maybe it's hardware...

It wasn't. A new install fixed everything.

> Have you had a lightning storm in the area recently?  

Yeah. That's common in Colorado.

> And is everything
> connected to the machine plugged into a power strip with a surge arrestor?

UPS. Where everything coming in goes to the battery, and everything out is 
generated by an oscillator and an amplifier.

> Reportedly, most lightning damage to computer gear is due to transients
> entering through the telephone line, rather than from the power line.

Nope again. This is a WiFi box.

-- 
Glenn English

Re: bewildering boot problem -- a little more info

[tomas]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

On Sun, Oct 04, 2015 at 06:14:44PM -0600, Glenn English wrote:
> I disabled XDM (sure is nice to have text config files) and rebooted. It came 
> up at a regular login prompt, and no error messages. startx took a *very* 
> long time -- said it couldn't find the hostname (something like that) -- but 
> it eventually started XFCE. And the messages were back.
> 
> Something I forgot to mention: When XFCE starts and I open a terminal, the 
> terminal comes up right away, but it takes  a long time (30+ seconds, it 
> feels like) to display the prompt.

All those 30-ish seconds timeout and the error message you mention
above smell of something trying to resolve a host name, failing
and giving up (timeout). Although I can't, for the life of me, imagine
what DNS lookups might be involved in the starting of a term.

To poke a bit in the dark, try (in a console) "ping localhost". What
happens? Then try "ssh @localhost". But I'm poking in
the dark.

Regards
- -- tomás
-BEGIN PGP SIGNATURE-
Version: GnuPG v1.4.12 (GNU/Linux)

iEYEARECAAYFAlYSIQ0ACgkQBcgs9XrR2kZyEQCcCi00QkAOR9TyRc8Ut8834eGP
9YIAoIA9lMqPVjJQAC5ErN902IV6zWVH
=W1jA
-END PGP SIGNATURE-

Re: bewildering boot problem -- a little more info

[Glenn English]

On Oct 5, 2015, at 1:04 AM,   wrote:

> All those 30-ish seconds timeout and the error message you mention
> above smell of something trying to resolve a host name, failing
> and giving up (timeout).

Yup. The timeouts do, anyway. The error messages from the kernel when I start 
XFCE never time out.

> Although I can't, for the life of me, imagine
> what DNS lookups might be involved in the starting of a term.

It's not just a term, it's somewhere in the X software. Most likely in XFCE. 

I don't understand the name lookup either, and the resolve config is set to 
look at /etc/hosts before hitting DNS. All the local hosts are in there. But 
I've seen the XFCE startup whine about not being to do DNS type things before.

> To poke a bit in the dark, try (in a console) "ping localhost". What
> happens?

Works great.

> Then try "ssh @localhost".

Hadn't though of trying ssh, but I bet it works to localhost. I'll try it in 
the morning.


I found those two things the kernel error messages are about 
(CAP_NET_ and the other CAP_ thing). They're in the kernel source 
code, in a file called something like capabilities.h. They're both #defines -- 
one is 15; the other is a different small integer (17 or 18, IIRC). Now I 
really have no idea what the boot is talking about. Sounds like it's telling me 
the kernel needs to be recompiled. 

They do have to do with setting capabilities in networking, though.

-- 
Glenn English

Re: bewildering boot problem

[rlharris]

In reply to Glenn English :

Inasmuch as you are running Xfce, is there any particular reason for
staying with Wheezy, which was the release which introduced the new Gnome?
 If not, the investment of two or three hours should have you up and
running in Jessie.

I have had opportunity to install Jessie with Xfce on several machines,
both amd64 and i386; they all have been running perfectly.

Sometimes operator results in a corrupted file, and sometimes a drive near
the end of its life can cause problems.  Besides, a fresh installation is
an opportunity for housecleaning.

Russ