Slitaz: Cannot install CUPS successfully

fopetesl · 03-23-2010, 12:45 PM

SLITAZ cooker (latest version)when I attempt to install CUPS

Code:

tazpkg get-install cups --force

cups-1.4.2

Installation of : cups-1.4.2
==================================
Copying cups-1.4.2... [ OK ]
Extracting cups-1.4.2... 3297 blocks
Extracting the pseudo fs... (lzma) 16018 blocks
Installing cups... [ OK ]
Removing all tmp files... [ OK ]
Starting cups deamon: Cupsd... [ Failed ]
==================================
cups (1.4.2) is installed.

I had a look for cups error log

Code:

tail /var/log/cups/errors.log
...No such file or directory

also

Code:

ps -A | grep cups
doesn't show cups
_but_
ps -A | grep cups*
does show "cups_install_log" for some reason.

When I attempt to start cups

Code:

Starting cups daemon: Cupsd... [Failed]
Child exited on signal 11!

I posted my problem on the SLITAZ Support Forum but so far no real help

rigor · 03-23-2010, 11:55 PM

It looks like you've installed cups, but just can't run the daemon. After you installed cups, did you configure it and define a printer to manage?

fopetesl · 03-24-2010, 04:56 AM

Quote:

Originally Posted by kakaka

It looks like you've installed cups, but just can't run the daemon. After you installed cups, did you configure it and define a printer to manage?

AH! You got me there

Configure? Are there instructions available?

rigor · 03-24-2010, 04:43 PM

AFAIK, unless it has changed fairly recently, there is a desire to have someone write a manual on printing in SLITAZ.

Bur I may missed something. Please provide the output of ps with the complete command line where you find the reference to "cups_install_log" in the ps output. Also, when you installed CUPS was your printer turned on and connected to the machine?

fopetesl · 03-25-2010, 03:54 AM

Quote:

Originally Posted by kakaka

AFAIK, unless it has changed fairly recently, there is a desire to have someone write a manual on printing in SLITAZ.

Wholeheartedly agree.

Quote:

Originally Posted by kakaka

Bur I may missed something. Please provide the output of ps with the complete command line where you find the reference to "cups_install_log" in the ps output. Also, when you installed CUPS was your printer turned on and connected to the machine?

Well you did ask

...

Code:

PID   USER     COMMAND
    1 root     init       
    2 root     [kthreadd]
    3 root     [migration/0]
    4 root     [ksoftirqd/0]
    5 root     [events/0]
    6 root     [khelper]
   12 root     [async/mgr]
  134 root     [kblockd/0]
  136 root     [kacpid]
  137 root     [kacpi_notify]
  202 root     [ata/0]
  203 root     [ata_aux]
  204 root     [ksuspend_usbd]
  210 root     [khubd]
  213 root     [kseriod]
  256 root     [pdflush]
  257 root     [pdflush]
  258 root     [kswapd0]
  259 root     [aio/0]
  260 root     [nfsiod]
  262 root     [xfs_mru_cache]
  263 root     [xfslogd/0]
  264 root     [xfsdatad/0]
  265 root     [xfsconvertd/0]
  266 root     [crypto/0]
  466 root     [scsi_eh_0]
  469 root     [scsi_eh_1]
  558 root     [kpsmoused]
  563 root     [kondemand/0]
  597 root     [usbhid_resumer]
  601 root     [rpciod/0]
  622 root     [kjournald]
  667 root     /sbin/syslogd -s 60 
  669 root     /sbin/klogd 
  671 root     /sbin/udevd acpi 
  939 root     /sbin/udevd acpi 
  944 root     /sbin/udevd acpi 
 1094 root     /sbin/udhcpc -b -T 1 -A 12 -i eth1 -p /var/run/udhcpc.eth1.pid 
 1107 messageb /usr/bin/dbus-daemon --system 
 1113 haldaemo /usr/sbin/hald --daemon=yes 
 1114 root     hald-runner 
 1120 root     hald-addon-input: Listening on /dev/event1
 1126 root     hald-addon-storage: polling /dev/sr0 (every 2 sec)
 1130 haldaemo hald-addon-acpi: listening on acpi kernel interface /proc/acpi/event
 1134 root     /usr/bin/slim -d 
 1138 root     /usr/bin/Xorg -auth /var/run/slim.auth vt07 
 1144 root     /sbin/getty 38400 tty1 
 1145 root     /sbin/getty 38400 tty2 
 1148 root     /sbin/getty 38400 tty3 
 1151 root     /sbin/getty 38400 tty4 
 1154 root     /sbin/getty 38400 tty5 
 1155 root     /sbin/getty 38400 tty6 
 1163 almeter  /usr/bin/openbox 
 1176 almeter  pcmanfm -d 
 1177 almeter  lxpanel 
 1180 almeter  dbus-launch --exit-with-session 
 1181 almeter  /usr/bin/dbus-daemon --fork --print-pid 5 --print-address 7 --session 
 1184 almeter  /usr/lib/gamin/gam_server 
 1186 almeter  /usr/lib/menu-cache/menu-cached 
 1189 almeter  /bin/sh /usr/bin/firefox 
 1201 almeter  /bin/sh /usr/lib/firefox-3.5.7/run-mozilla.sh /usr/lib/firefox-3.5.7/firefox-bin 
 1205 almeter  /usr/lib/firefox-3.5.7/firefox-bin 
 1261 almeter  xterm -ls 
 1262 almeter  -sh 
 1265 root     sh 
 1271 root     ps

Printer not plugged in yet! This is important? Ubuntu (nice but slo.o.o.o.wer) doesn't object.

So I poked around the workshop and plugged in an HP Deskjet D1660.
No change.

Cups still reports the same error.

rigor · 03-25-2010, 06:46 AM

OK, my bad, I need to be more specific.

I was looking for the cups_install_log you mentioned in the output of the ps command.

But grep cups will already find cups ( lower case ) in any sequence of characters of which it is part. If cups_install_log had been in the ps output, it should have already found it, unless the ps output changed between the first grep and the second.

Also, if you want to use a pattern with grep, then you want to escape or quote the pattern so the shell doesn't evaluate it, and you probably want to use egrep.

Very likely you were in a directory with some file named cups_install_log or with a file that contained the phrase cups_install_log. The shell matched cups* in the directory you were in, so after grep finished grep'ing it's own stdin provided by the output of the ps command, it then started grep'ing through files whose names matched cups* because the matched names became part of grep's command line.

What I was trying to suggest before is this:

1) Uninstall cups.
2) Make sure you have a printer properly connected to the machine and turned on.
3) As root, re-install cups.
4) Get into the same directory you were when you saw the phrase cups_install_log come out of the grep.

Then, look for a file with a name like cups_install_log or some sort of cups install log. cups_install_log could have been a file name. Or there could have been a file with a name that matched the pattern cups* and the file may have contained the phrase cups_install_log.

Hopefully the cups install log file, whatever it is named, will tell you what's happening.

fopetesl · 03-25-2010, 10:04 AM

OK. My bad this time!
I created cups_install_log over a week ago to try and capture the output when starting the daemon.

I also went through all the config options I could google without any change.

So my last attempt is to

Code:

~# strace /etc/init.d/cupsd start
execve("/etc/init.d/cupsd", ["/etc/init.d/cupsd", "start"], [/* 22 vars */]) = 0
brk(0)                                  = 0x80d0000
access("/etc/ld.so.preload", R_OK)      = -1 ENOENT (No such file or directory)
open("/usr/lib/tls/i686/sse2/libm.so.6", O_RDONLY) = -1 ENOENT (No such file or directory)
stat64("/usr/lib/tls/i686/sse2", 0xbfed467c) = -1 ENOENT (No such file or directory)
open("/usr/lib/tls/i686/libm.so.6", O_RDONLY) = -1 ENOENT (No such file or directory)
stat64("/usr/lib/tls/i686", 0xbfed467c) = -1 ENOENT (No such file or directory)
open("/usr/lib/tls/sse2/libm.so.6", O_RDONLY) = -1 ENOENT (No such file or directory)
stat64("/usr/lib/tls/sse2", 0xbfed467c) = -1 ENOENT (No such file or directory)
open("/usr/lib/tls/libm.so.6", O_RDONLY) = -1 ENOENT (No such file or directory)
stat64("/usr/lib/tls", 0xbfed467c)      = -1 ENOENT (No such file or directory)
open("/usr/lib/i686/sse2/libm.so.6", O_RDONLY) = -1 ENOENT (No such file or directory)
stat64("/usr/lib/i686/sse2", 0xbfed467c) = -1 ENOENT (No such file or directory)
open("/usr/lib/i686/libm.so.6", O_RDONLY) = -1 ENOENT (No such file or directory)
stat64("/usr/lib/i686", 0xbfed467c)     = -1 ENOENT (No such file or directory)
open("/usr/lib/sse2/libm.so.6", O_RDONLY) = -1 ENOENT (No such file or directory)
stat64("/usr/lib/sse2", 0xbfed467c)     = -1 ENOENT (No such file or directory)
open("/usr/lib/libm.so.6", O_RDONLY)    = -1 ENOENT (No such file or directory)
stat64("/usr/lib", {st_mode=S_IFDIR|0755, st_size=20480, ...}) = 0
open("/lib/tls/i686/sse2/libm.so.6", O_RDONLY) = -1 ENOENT (No such file or directory)
stat64("/lib/tls/i686/sse2", 0xbfed467c) = -1 ENOENT (No such file or directory)
open("/lib/tls/i686/libm.so.6", O_RDONLY) = -1 ENOENT (No such file or directory)
stat64("/lib/tls/i686", 0xbfed467c)     = -1 ENOENT (No such file or directory)
open("/lib/tls/sse2/libm.so.6", O_RDONLY) = -1 ENOENT (No such file or directory)
stat64("/lib/tls/sse2", 0xbfed467c)     = -1 ENOENT (No such file or directory)
open("/lib/tls/libm.so.6", O_RDONLY)    = -1 ENOENT (No such file or directory)
stat64("/lib/tls", 0xbfed467c)          = -1 ENOENT (No such file or directory)
open("/lib/i686/sse2/libm.so.6", O_RDONLY) = -1 ENOENT (No such file or directory)
stat64("/lib/i686/sse2", 0xbfed467c)    = -1 ENOENT (No such file or directory)
open("/lib/i686/libm.so.6", O_RDONLY)   = -1 ENOENT (No such file or directory)
stat64("/lib/i686", 0xbfed467c)         = -1 ENOENT (No such file or directory)
open("/lib/sse2/libm.so.6", O_RDONLY)   = -1 ENOENT (No such file or directory)
stat64("/lib/sse2", 0xbfed467c)         = -1 ENOENT (No such file or directory)
open("/lib/libm.so.6", O_RDONLY)        = 3

========== only 1st part of output =================
which looks like a bunch of dependencies missing?
Note there is no prompt regarding depends using the software installer - all I see is a tick box with "auto install depends" and "get-install" which suggests depends are automatically installed.
..or from the CLI are there any prompts for depends.

rigor · 03-25-2010, 12:05 PM

I was bouncing all over my system trying to find my cups_install_log, very puzzled not to find any reference to it, even inside the cups package itself. :-)

As you say, the installer should have checked all the dependencies. What you are seeing is the way processes search for dynamic components.

Sadly it often involves a hard-coded list of places to look. Sometimes many of those don't even exist on the system. That seems rather odd when you consider that there are standard mechanisms on Linux for specifying which directories are used on the system to hold such files.

You can see the zero return code, meaning success, where it found the /usr/lib directory.

That sequence will probably continue on, and repeat looking for different files, until everything it wants to find, is found, or it fails to find something, anywhere it looks.

Besides, AFAIK the signal a child process of cupsd caught, the one that you first illustrated, is a SEGV. Usually that would mean something more serious happened.

So if you make sure you strace the child process too, usually the -f flag does that, as well as put in enough options to bump up the level of details about as high as it goes, you might be able to see towards the end of the file, what the problem actually is.

fopetesl · 03-27-2010, 05:50 AM

For myself this is mostly a foreign language. These are the last few lines of # strace -f /etc/init.d/cupsd start

Code:

pid  1309] open("/var/cache/cups/job.cache", O_RDONLY|O_LARGEFILE) = -1 ENOENT (No such file or directory)
[pid  1309] open("/etc/cups/subscriptions.conf", O_RDONLY|O_LARGEFILE) = -1 ENOENT (No such file or directory)
[pid  1309] open("/var/spool/cups/tmp", O_RDONLY|O_NONBLOCK|O_LARGEFILE|O_DIRECTORY|O_CLOEXEC) = 4
[pid  1309] getdents64(4, /* 2 entries */, 32768) = 48
[pid  1309] getdents64(4, /* 0 entries */, 32768) = 0
[pid  1309] close(4)                    = 0
[pid  1309] access("/etc/gcrypt/fips_enabled", F_OK) = -1 ENOENT (No such file or directory)
[pid  1309] open("/proc/sys/crypto/fips_enabled", O_RDONLY) = -1 ENOENT (No such file or directory)
[pid  1309] brk(0xb80f6000)             = 0xb80f6000
[pid  1309] access("/dev/random", R_OK) = 0
[pid  1309] access("/dev/urandom", R_OK) = 0
[pid  1309] time(NULL)                  = 1269686550
[pid  1309] --- SIGSEGV (Segmentation fault) @ 0 (0) ---
Process 1309 detached
[pid  1308] <... nanosleep resumed> {0, 173226535}) = ? ERESTART_RESTARTBLOCK (To be restarted)
[pid  1308] --- SIGCHLD (Child exited) @ 0 (0) ---
[pid  1308] sigreturn()                 = ? (mask now [])
[pid  1308] wait4(-1, [{WIFSIGNALED(s) && WTERMSIG(s) == SIGSEGV}], 0, NULL) = 1309
[pid  1308] write(2, "cupsd: Child exited on signal 11"..., 34cupsd: Child exited on signal 11!
) = 34
[pid  1308] exit_group(3)               = ?
Process 1306 resumed
Process 1308 detached
<... waitpid resumed> [{WIFEXITED(s) && WEXITSTATUS(s) == 3}], 0) = 1308
--- SIGCHLD (Child exited) @ 0 (0) ---
fcntl64(1, F_GETFL)                     = 0x8002 (flags O_RDWR|O_LARGEFILE)
write(1, "\33[70G[ ", 7                                              [ )                = 7
fcntl64(1, F_GETFL)                     = 0x8002 (flags O_RDWR|O_LARGEFILE)
write(1, "\33[1;31mFailed", 13Failed)         = 13
fcntl64(1, F_GETFL)                     = 0x8002 (flags O_RDWR|O_LARGEFILE)
write(1, "\33[0;39m ]\n", 10 ]
)           = 10
exit_group(0)                           = ?

Something to do with [pid 1308] write(2, "cupsd: Child exited on signal 11"..., 34cupsd: Child exited on signal 11!
) = 34?
But what??

I also did an # strace -fFv /etc/init.d/cupsd start but it didn't seem to show any more indication of the problem around the end of [pid 1308] where the SIGSEGV ocurred.

rigor · 03-27-2010, 04:58 PM

The write you mentioned is just writing the error message to the error file.

I probably don't have enough to go on, so this could be wrong. But the value returned from the time() function call seems to represent a date in 2010. It almost looks like the child process did a successful call to the time() function, then ran out of stack space to store the time value. If so, you might just need to bump up your stack space, if possible. That generally shouldn't be needed, but SLITAZ is supposed to be small, so it's not impossible I suppose.

Could you please do these things. Just for grins, try to uninstall cups, make sure your printer is connected to the machine and the printer turned on, then try to reinstall cups while capturing the output of the install procedure into a log and upload the log to this thread as an attachment.

Then if cupsd still fails to start, perhaps you could run this command:

Code:

strace  -Ff  -ff strace_procs  -ttt  -v  -xx  /etc/init.d/cupsd start  >  strace_full.log  2>&1

and upload the strace_full.log and strace_procs.* files to this thread as attachments.

Please tell us if you know, which shell ( bash, sh, etc. ) you are using when running as root, and even if you don't know which shell you are using, as root please run this command:

Code:

ulimit -a

and paste it's output into this thread.

Thanks!

fopetesl · 03-28-2010, 09:34 AM

OK. But how to capture the install messages?
# install > log only saves some of the result.

Anxious to sort this

rigor · 03-28-2010, 01:00 PM

Quote:

Originally Posted by fopetesl

OK. But how to capture the install messages?
# install > log only saves some of the result.

Anxious to sort this

Code:

install  >  log  2>&1

should get them all.

fopetesl · 03-29-2010, 04:57 AM

Code:

strace  -Ff  -ff strace_procs  -ttt  -v  -xx  /etc/init.d/cupsd start  >  strace_full.log  2>&1

fails:

Code:

# cat strace_full.log 
strace: strace_procs: command not found

also

Code:

# tazpkg get-install strace_procs
Unable to find : strace_procs in the mirrored packages list.
Unable to find : get-strace_procs in the mirrored packages list.

Doesn't get any easier

rigor · 03-29-2010, 03:53 PM

OOOPS! My bad! Big BOOBOO!! Sorry, I was thinking at the same time, both about capturing the output from the install, as well as doing the strace, and cross-pollenating concepts between the two! Not enough caffeine! :-O

This should be what you want:

Code:

strace  -Ff  -ff -ttt  -v  -xx  -o strace_log  /etc/init.d/cupsd start

The -ff is supposed to put the strace for each process in it's own file with the process ID of the file appended to the file name provided. So if one of the processes ran with a process ID of 28882 you'd get a file named strace_log.28882 with the strace output of tracing that process.

That command form works on my system if I try to run something that I know will launch several processes, I get several strace output files. But your system maybe different. For example, I'm including the capital F option because it was supported on older systems and I believe you were using it in your example. In later systems it is considered deprecated/non-functional. But in later systems, the command is still aware of the option, so it doesn't complain if it is used.

fopetesl · 03-30-2010, 04:43 AM

1st I cleaned the cache(s):

Code:

r# tazpkg clean-cache

Clean cache : /var/cache/tazpkg
================================================================================
Cleaning cache directory...                                          [ OK ]
================================================================================
21 file(s) removed from cache.

(after removing cups), then (printer plugged in and switchd ON):

Code:

# tazpkg get-install cups

cups-1.4.2
Connecting to mirror.slitaz.org (94.23.60.116:80)
cups-1.4.2.tazpkg    100% |*************************************************************************************************************************************|  1648k 00:00:00 ETA

Installation of : cups-1.4.2
================================================================================
Copying cups-1.4.2...                                                [ OK ]
Extracting cups-1.4.2... 3297 blocks
Extracting the pseudo fs... (lzma) 16018 blocks
Installing cups...                                                   [ OK ]
Removing all tmp files...                                            [ OK ]
Starting cups deamon: Cupsd... cupsd: Child exited on signal 11!
                                                                     [ Failed ]
================================================================================
cups (1.4.2) is installed.

Ummm. Can only upload three (renamed to *.txt) files so I'll try to add as another Reply.
Now:

Code:

# ulimit -a
time(seconds)        unlimited
file(blocks)         unlimited
data(kb)             65536
stack(kb)            8192
coredump(blocks)     0
memory(kb)           65536
locked memory(kb)    64
process              3575
nofiles              1024
vmemory(kb)          unlimited
locks                unlimited

... more ...(?)