Slitaz: Cannot install CUPS successfully

fopetesl · 03-30-2010, 04:45 AM

Lets see if we can do this...
OK. Seemed to upload ..

Bated breath is the situation

rigor · 03-30-2010, 06:10 AM

Thanks!

It looks like it's trying to get more memory, at times. It's also doing what's known as memory mapping.

After you login as root, but before you try running cupsd, you might try running this in the root shell:

Code:

ulimit  -v unlimited  -l unlimited  -m unlimited  -s unlimited

you may have to double check the options to ulimit on your system.
The idea is to make as much memory as possible available to the
cupsd.

Then try to start cupsd.

Also, you might want to check for any local system e-mail to root.

fopetesl · 03-30-2010, 06:38 AM

Nope. Still same.

Code:

# ulimit -l
unlimited

Also I note but maybe not relevant?

Code:

Scanning localhost ports 1 to 1024
 Port   Proto   State   Service
1024 closed, 0 open, 0 timed out (or blocked) ports

I don't seem to have an e_mail program on this box.
Any other way to check?

I looked at var/log/messages:

Code:

Mar 30 11:27:14 (none) user.info kernel: cupsd[4968]: segfault at 0 ip (null) sp bf84e400 error 4 in cupsd[40000000+52000]

I also note Slitaz V3.0 is now out. Any reason to try upgrading?

Though it would be more educational to dig out what is going screwy here first.

rigor · 03-30-2010, 10:53 PM

re - memory limits:

Please make sure that the command:

Code:

ulimit -a

shows all the various memory limits as unlimited.

For that matter, how much physical memory is in the machine?

re - mail:

Typically this would show you the local mailboxes, if any.

Code:

ls -a /var/spool/mail

but you might be able to double check the location with the command

Code:

echo $MAIL

to see to what directory it points.

re - ports:

Generally, cupsd wants to control port 631 at least. Do you have a firewall in place? Some such facilities can be used to prevent inside access. Although often the main focus would be to prevent outside access. But you might want to double check. Do you have iptables or eptables installed, or any sort of firewall software?

re - upgrading:

I don't see any mention in the release notes that cups now installs, if that's what you mean by a reason for upgrading.

When I upgrade, if possible, I usually try to arrange things to that I can multi-boot, the old release side by side with the new release, in case there is anything wrong with the new release, and I decide I want to keep running the old. Or just in case I want to double check the way things worked in the old release, I can boot the old release.

fopetesl · 03-31-2010, 03:28 AM

Firstly I ran

Code:

ulimit  -v unlimited  -l unlimited  -m unlimited  -s unlimited

and it baulked:

Code:

:~# ulimit  -v unlimited  -l unlimited  -m unlimited  -s unlimited    
sh: ulimit: too many arguments

so I ran each ulimit command individually:

Code:

:~# ulimit -v unlimited 
:~# ulimit -l unlimited 
:~# ulimit -m unlimited 
:~# ulimit -s unlimited 
:~# ulimit
unlimited

Ah! my typo

Code:

~# ulimit -a
time(seconds)        unlimited
file(blocks)         unlimited
data(kb)             65536
stack(kb)            unlimited
coredump(blocks)     0
memory(kb)           unlimited
locked memory(kb)    unlimited
process              3575
nofiles              1024
vmemory(kb)          unlimited
locks                unlimited

Code:

:~# ls -a /var/spool/mail
ls: /var/spool/mail: No such file or directory
~# echo $MAIL    
/var/mail/Peters
:~# cd /var/mail/Peters
sh: cd: can't cd to /var/mail/Peters
:~# ls -l /var/spool/Peters
ls: /var/spool/Peters: No such file or directory

Code:

:~# iptables
sh: iptables: not found
:~# eptables
sh: eptables: not found

Any other way to check if I have a Firewall installed?

Upgrade:

Ummm. Only to perhaps cure the problem? Though no one else has (so far) commented about install problems with cups with V2.0
If I can find time today I'll try V3.0 on another clean hard drive.

rigor · 03-31-2010, 04:51 AM

For ulimit, the remaining value I believe usually uses the d flag, as in:

Code:

ulimit -d unlimited

If you could get to the point where all the memory related items say unlimited, and then try starting cupsd, that would be nice.

Also, how much physical memory is on the machine?

The MAIL variable points to what can be a mail file if there is local mail for a particular User.

What you'd want to do as root is something like this to look at what else might be in that directory:

Code:

ls -al /var/mail

I saw a mention of mail in the trace files, and I wondered if maybe the cupsd might have mailed root, or some "system" user locally. If it had, the mail might be in that directory.

Sure, installing the new version of the OS might be worth a shot.

As to the firewall, the only things I know of on the list of packages for that distro that might implement a firewall are the *tables stuff.

Can you get a list of all packages you have installed so we could check if there might be something else that has the port closed?

fopetesl · 03-31-2010, 06:18 AM

Code:

:~# ulimit -v unlimited 
:~# ulimit -l unlimited 
:~# ulimit -m unlimited 
:~# ulimit -s unlimited 
:~# ulimit -d unlimited 
:~# ulimit -n unlimited 
sh: ulimit: error setting limit (Operation not permitted)
:~# ulimit -c unlimited 
:~# ulimit -p unlimited 
:~# ulimit -a
time(seconds)        unlimited
file(blocks)         unlimited
data(kb)             unlimited
stack(kb)            unlimited
coredump(blocks)     unlimited
memory(kb)           unlimited
locked memory(kb)    unlimited
process              unlimited
nofiles              1024
vmemory(kb)          unlimited
locks                unlimited

Stiil same error when attempting to run cupsd

512MB RAM installed.

So I looked at:

Code:

# ls -al /var/mail
ls: /var/mail: No such file or directory

...and...

Code:

# ls -l /var/spool
drwxr-xr-x    3 root     root         4096 Mar 25 12:40 cron
drwx--x---    3 root     nogroup      4096 Mar 30 11:27 cups

So I did a search for any mail directory of significance. Nothing.

As to installed programs all I can find is

Code:

e# cat /var/log/tazpkg.log 
2010-03-25 13:39:31 - Installed - mc (4.7.0.1) - 792fe3af5a9c3b8e1f48ae3b6b98b9c6
2010-03-25 13:40:11 - Installed - openssl (0.9.8l) - 9f3b703b954244c2120de1918d093bc5
2010-03-25 13:40:15 - Installed - libgpg-error (1.7) - 1740323af4b2ce859430dbc8933fb316
2010-03-25 13:40:15 - Installed - libgcrypt (1.4.4) - 41c51716524014c45af8e327c91bb992
2010-03-25 13:40:16 - Installed - readline (6.0) - 93597b1eea82eebe3e07c937cadd6b3e
2010-03-25 13:40:18 - Installed - libgnutls (2.8.3) - 33500aad7bbe287bb235e4fb4fdc8369
2010-03-25 13:40:19 - Installed - libtasn1 (2.3) - 5f3665f38b3be77514370d0bedb3e015
2010-03-25 13:40:23 - Installed - make (3.81) - 6e37b7395ffa82c65ef5630c2e27dec2
2010-03-25 13:40:23 - Installed - gmp (4.2.2) - 549b48e2af6470f3babfcab6a407900c
2010-03-25 13:40:25 - Installed - libtool (2.2.2) - 4097b0e4f2f22d1532911683b43a9477
2010-03-25 13:40:27 - Installed - guile (1.8.7) - 0ccb07cdcc04f5ac4cd7042ba8e1abae
2010-03-25 13:40:29 - Installed - gnutls (2.8.3) - adda127d8659530921f32e7d322f8192
33500aad7bbe287bb235e4fb4fdc8369
2010-03-25 13:40:32 - Installed - libkrb5support (1.6.3) - 2ddb2c08686701d35495fbbe2991679f
2010-03-25 13:40:34 - Installed - libcomerr3 (1.6.3) - 93c67e8178bd932f328d22b4df8a169d
2010-03-25 13:40:34 - Installed - libkrb5 (1.6.3) - 2b87fd2c4773048bff641ad5e47f9262
2010-03-25 13:40:36 - Installed - libusb-compat (0.1.0) - 3a8b21e39c12a1cc2566c78925d39a90
2010-03-25 13:40:36 - Installed - libusb (1.0.0) - 9149ebe1d02153c70a7ebe1558b65910
2010-03-25 13:40:40 - Installed - cups (1.4.2) - 1317345ed07c0797e4b95e2cac67ca2d
2010-03-25 13:57:19 - Installed - strace (4.5.19) - 918b64f955dee88e945710b753dbc685
2010-03-25 14:50:53 - Installed - cups (1.4.2) - 1317345ed07c0797e4b95e2cac67ca2d
2010-03-25 14:52:08 - Reconfigured - cups (1.4.2)
2010-03-25 14:52:31 - Removed - cups (1.4.2)
2010-03-25 14:52:44 - Installed - busybox (1.12.0) - be0daef778c8009535fd0913908671bc
2010-03-25 14:54:58 - Installed - cups (1.4.2) - 1317345ed07c0797e4b95e2cac67ca2d
2010-03-27 11:01:40 - Installed - lighttpd (1.4.25) - 3fe435e9ec7ce3e8d1d3a878a9381d2e
2010-03-29 09:48:50 - Removed - cups (1.4.2)
2010-03-29 09:51:06 - Installed - cups (1.4.2) - 1317345ed07c0797e4b95e2cac67ca2d
2010-03-30 09:26:57 - Removed - cups (1.4.2)
2010-03-30 09:27:07 - Installed - busybox (1.12.0) - be0daef778c8009535fd0913908671bc
2010-03-30 09:27:27 - Installed - cups (1.4.2) - 1317345ed07c0797e4b95e2cac67ca2d
2010-03-30 11:27:14 - Installed - cups (1.4.2) - 1317345ed07c0797e4b95e2cac67ca2d

I had a poke around the system and found these:

Code:

# tail /var/log/messages
Mar 31 10:30:58 (none) user.info kernel: cupsd[1314]: segfault at 0 ip (null) sp bff67660 error 4 in cupsd[40000000+52000]

__and an old

Code:

# cat /var/log/cups/error_log 
E [27/Mar/2010:11:35:54 +0000] Hostname lookup for "hostname" failed!
E [27/Mar/2010:11:35:54 +0000] Bad Listen address hostname:80 at line 17.

but it's a few days old and we've tried running cupsd several times since then

I had a look into /etc/cups/cups.conf to (re)checl the "Listen" ports:

Code:

Listen localhost:631
Listen /var/run/cups/cups.sock

but the cups.sock file doesn't exist.

Upgrade. I tried V3.0 fresh install on spare hard drive. No change - same error. and Midori crashed in the middle of replying. I'm still with V2.0.

rigor · 03-31-2010, 06:25 PM

Thanks. This new information opens a variety of possibilities.

You realize of course, that if this problem gets solved, you could potentially write the much needed manual on installing cups in the distro!

OK, so it doesn't appear to be a memory problem.

The install log looks a bit inconsistent. It almost looks as though it let you install cups when cups was already installed and produced different results, for different re-installs. It seems unknown as to whether or not it should consider cups to be installed properly, the way it seems to, even though cupsd won't start. Also, are those all the that packages that are installed, or just packages that have been installed recently?

So you are saying that when you try to run cupsd currently, you don't get any current log messages about the hostname lookup problem, is that right? Your machine has a host name within Linux with an associated IP address, even if it's a default host name and IP address, right? For example, if you run the command:

Code:

hostname

you get some name displayed. Then if you take the name displayed by the hostname command, say for example the name is linux which is a default on some distros, and you then use the grep command to look for that host name in your /etc/hosts file, for example:

Code:

grep linux /etc/hosts

you'll find an ip address in /etc/hosts associated with whatever host name the hostname command outputs, yes?

Would you please try to attach the cups config file and log files to this thread.

Thanks.

fopetesl · 04-01-2010, 03:38 AM

Quote:

You realize of course, that if this problem gets solved, you could potentially write the much needed manual on installing cups in the distro!

Like I have time. Already.

But if we get it up and running I'll find the time!

The install log seems to be complete though I have no way to be 100% certain. It does include the programs that Slitaz installed automagically when installing to HDD initially so probably true.
The multiple installs of cups were varieties of install-from-cache, remove-then-reinstall and clean-the-download-cache-and-do-a-fresh-install. The busybox "(re)install"s were a response to "do you want to repair/reinstall programs modified by cups install?".

Code:

# hostname
Peters
root@Peters:/home/peters# grep Peters /etc/hosts 
root@Peters:/home/peters# grep slitaz /etc/hosts       
127.0.0.1      localhost slitaz

So I added the line to /etc/hosts: 127.0.0.1 localhost Peters.

Well, that didn't work!

Code:

root@Peters:/etc# /etc/init.d/cupsd start
Starting cups deamon: Cupsd... cupsd: Child exited on signal 11!
                                                                     [ Failed ]

Umm. Sorry. Which log file(s) do you want to see?
/var/log/cups/error.log hasn't been updated since 27 March.
/var/log/messages has only the same as previously - cupsd segfault.

rigor · 04-01-2010, 04:41 AM

as to the log files I was just trying to double check that it was no definitely no longer complaining about hostname.

OK, I think it's time to focus on the version of the distro you're using.

When you started this thread you referred to SLITAZ cooker. So you are using the latest "cooking" version of 2.x, not the latest stable version of 2.x - is that right? What "flavor" are you using? Another to put that is, what were the names of the ISO's you installed? Or how did you go about the installation?

fopetesl · 04-01-2010, 05:58 AM

Quote:

Originally Posted by kakaka

When you started this thread you referred to SLITAZ cooker. So you are using the latest "cooking" version of 2.x, not the latest stable version of 2.x - is that right? What "flavor" are you using? Another to put that is, what were the names of the ISO's you installed? Or how did you go about the installation?

I have it loaded on my BIG computer as Slitaz-cooking-firefox.iso.
However. Please note my earlier comment about trying Slitaz-V3.0-stable.

Nice idea but the logic doesn't hold up for distro specific. Other people have complained only of specific printer problems so maybe it's a hardware issue? - printer is plugged in and switched on though I cannot see why this may be significant.
Previously other distros, mostly Ubuntu, load cups without a printer connected and either prompt at the GUI or install through localhost:631.

Installation was via the LIVE interface since there doesn't seem to be an alternative.
Also I had to wipe the disk and create an ext3 partition (of the whole HDD) for installation, i.e. the partition has to exist or Slitaz won't install. Same for V3.0 BTW.

Actually I note on the Slitaz downloads page http://www.slitaz.org/en/get/index.html#stable (stable & cooking are all on the same page) that there are non-LIVE iso images. Should I try one of these I wonder.

fopetesl · 04-01-2010, 11:45 AM

I have just completed a test using an older mobo, a Gigabyte G-8 board with 256MB SDRAM and Radeon AGP card. Same HDD.

cups loads and runs with no error!
And. I can connect using localhost:631.

Howto dig out THIS problem?

Could not get internet - ifconfig shows only eth2 but will not get a dhcp lease. So no more info.

Ubuntu versions load cups OK on this board with both 5.10 (Breezy) and 9.10 (Karmic) without a problem.

I see that there is a later version of cups 1.4.3 instead of this version 1.4.2.

Safe to download and (how to?) compile?

rigor · 04-01-2010, 02:34 PM

Quote:

Distro specific? Ummm....
[...]
I have it loaded on my BIG computer as Slitaz-cooking-firefox.iso.
However. Please note my earlier comment about trying Slitaz-V3.0-stable.

Nice idea but the logic doesn't hold up for distro specific. Other people have complained only of specific printer problems so maybe it's a hardware issue? - printer is plugged in and switched on though I cannot see why this may be significant.
Previously other distros, mostly Ubuntu, load cups without a printer connected and either prompt at the GUI or install through localhost:631.

Installation was via the LIVE interface since there doesn't seem to be an alternative. Also I had to wipe the disk and create an ext3 partition (of the whole HDD) for installation, i.e. the partition has to exist or Slitaz won't install. Same for V3.0 BTW.

Actually I note on the Slitaz downloads page http://www.slitaz.org/en/get/index.html#stable (stable & cooking are all on the same page) that there are non-LIVE iso images. Should I try one of these I wonder.

Actually, I wasn't thinking it was a problem with SLITAZ specifically. I was wondering if it was a problem with a 2.x "cooking"/un-stable version versus the 2.x stable version. Also, there are supposedly different "flavors" of SLITAZ. Even though they don't seem to make it easy on the SLITAZ site to get older releases once the new release is out, some of the mirrors have archived older versions. Also, when I downloaded the 2.0 packages ISO from ibiblio the MD5 sum was wrong. There could be something wrong with the ISO, or they might just have put up the wrong MD5 sum. I wondered if the md5sum was correct for what you had. I also wondered if you had a "cooking" version of 2.x I was going to suggest you try the latest stable version of 2.x from a mirror, one with a valid md5sum.

rigor · 04-01-2010, 03:05 PM

Quote:

Originally Posted by fopetesl

I have just completed a test using an older mobo, a Gigabyte G-8 board with 256MB SDRAM and Radeon AGP card. Same HDD.

cups loads and runs with no error!
And. I can connect using localhost:631.

Howto dig out THIS problem?

Could not get internet - ifconfig shows only eth2 but will not get a dhcp lease. So no more info.

Ubuntu versions load cups OK on this board with both 5.10 (Breezy) and 9.10 (Karmic) without a problem.

I see that there is a later version of cups 1.4.3 instead of this version 1.4.2.

Safe to download and (how to?) compile?

OH! I missed the latest installment when I responded. Cool! And, Freakie! Cups having problems with a motherboard?

It was my understanding that the SLITAZ folks changed a lot of things to keep the system very very very small. I wasn't sure if you were using a SLITAZ modified cups that might have other requirements like that the printer be connected so it could try to configure the printer driver, or what. Since what seemed the most likely source of the problem didn't seem to help, I was just trying look at anything, however unlikely.

As to the new cups, if the problem with cups seems like it was something with the motherboard, it might worth it to try the newer version. You could also go back to the one that works if the new one doesn't. I can't comment on the newest cups specifically, because I'm still using 1.3. I've also never had to compile cups.

As to eth2, I take it that eth2 is the interface you want to use, it's up and running, properly configured, all the proper routing is set up. Do you know if it's even sending out packets? I usually keep an old true hub around, not a LAN switch, that I can connect to a machine when I'm troubleshooting network issues. That way I can use another machine to eavesdrop on the conversations of the computer under test. Do you have lights on the NIC that will tell you whether or not it's trying to send packets? Or something you can connect to check?

fopetesl · 04-02-2010, 07:39 AM

The eth2 problem isn't. Yet.
For some reason ifconfig didn't show either eth0 or eth1 on the 'old' mobo, only eth2. And dummy0, local, etc.
And although I set it up to dhcp at boot which seemed to work I couldn't ping or connect.
Seems that Slitaz doesn't fully recognise eth2

However, next step.
Downloaded and tar'red cups 1.4.3.
./config throws a wobbly. Has me baffled .. see the log file.
Doesn't look like depends are missing.