SEO

Hello to everyone,
I've got two sun machines (sun os 5.8), but one of them does not work
properly. First of all I have to say that accounts are based on a NIS
server (linux debian 3), which also exports home directories thru NFS.
Now, connecting to the two suns as normal user, I cannot list EVERY
directory thru ls, while on the other machine I can. Doing ls as root
everytime works well, while as normal users does not. Furthermore, even
if I cannot see the filesystem on a machine, I can write and read on it,
so that the couple:
echo "hello" > file.txt
cat file.txt
works properly, while the
ls -al
does not. I repeat that this happens only on one machine, while the
other works well. Firstly I thought as a NIS/NFS problem, but I cannot
list each directory, even if not mounted by NFS. Thus:
ls /
does not work.
In a second time I thought to a shell problem, but changing the current
shell (bash) to another (csh,ksh,sh) does not work.
Then I thought to an ls problem (machine compromised?), but the two ls
commands seems identical in both the machines, and event the ldd command
for ls reports the same libraries.
I've tried an "echo *" and it works, so I'm more convinced that's an ls
problem, but even copying the ls command from the machine where it works
to the broken one does not produce a result.
And now I don't know what to do to resume the broken machine.

Any idea?
Please help me.

Luca


--
Posted via Mailgate.ORG Server - http://www.Mailgate.ORG

In article <1509d1957db095b523a7d624a84a269c.60620@mygate.mai lgate.org>, "Luca Ferrari" <fluca1978@libero.it> writes:
> Hello to everyone,
> I've got two sun machines (sun os 5.8), but one of them does not work
> properly. First of all I have to say that accounts are based on a NIS
> server (linux debian 3), which also exports home directories thru NFS.
> Now, connecting to the two suns as normal user, I cannot list EVERY
> directory thru ls, while on the other machine I can. Doing ls as root
> everytime works well, while as normal users does not. Furthermore, even
> if I cannot see the filesystem on a machine, I can write and read on it,
> so that the couple:
> echo "hello" > file.txt
> cat file.txt
> works properly, while the
> ls -al
> does not. I repeat that this happens only on one machine, while the
> other works well. Firstly I thought as a NIS/NFS problem, but I cannot
> list each directory, even if not mounted by NFS. Thus:
> ls /
> does not work.
> In a second time I thought to a shell problem, but changing the current
> shell (bash) to another (csh,ksh,sh) does not work.
> Then I thought to an ls problem (machine compromised?), but the two ls
> commands seems identical in both the machines, and event the ldd command
> for ls reports the same libraries.
> I've tried an "echo *" and it works, so I'm more convinced that's an ls
> problem, but even copying the ls command from the machine where it works
> to the broken one does not produce a result.
> And now I don't know what to do to resume the broken machine.
>
> Any idea?
> Please help me.
>
> Luca
>

Things to try...

Network problem?
ping -s server 1400 10
mkfile 10m 10m

ls aliased to a bad command?
\ls -al

PATH component on a dead NFS server?
/bin/ls -al

Which mount options?
mount | grep $PWD
df .

Why the shell cannot be changed?
/bin/sh
should start an sh (exit to return to previous shell)
exec /bin/sh
should replace current shell by sh

System error message?
cat /var/adm/messages

--
Michael Tosch
IT Specialist
HP Managed Services Germany
Phone +49 2407 575 313

"Michael Tosch" <eedmit@NO.eed.SPAM.ericsson.PLS.se> wrote in message
news:bv8ina$ch4$1@aken.eed.ericsson.se


>
> Things to try...
>
> Network problem?
> ping -s server 1400 10
> mkfile 10m 10m
>

The NFS/NIS server works well, is on line and export well username and
password. Furthermore
I can write a file on the mounted filesystem and I can see it on the
file server
(i.e., the write works) but cannot list. I can see also if I do 2echo
file".


> ls aliased to a bad command?
> \ls -al

No, there are not strange alias, just ls -F. Nevertheless even executing
the
full qualified command /bin/ls does not work.


> PATH component on a dead NFS server?
> /bin/ls -al


No.

> Which mount options?
> mount | grep $PWD
> df .
>


Here they are:
bash-2.03$ mount | grep $PWD
/home/luca on lx01:/export/SunOS-sun4u/home/luca
remote/read/write/setuid/dev=3700028 on Wed Jan 28 16:56:09 2004

bash-2.03$ df .
/home/luca (lx01:/export/SunOS-sun4u/home/luca): 3559336 blocks
2862348 files




> Why the shell cannot be changed?
> /bin/sh
> should start an sh (exit to return to previous shell)
> exec /bin/sh
> should replace current shell by sh

Sorry, maybe I misdetailed the problem: I can of course change shell,
but even in
a different shell I cannot get the ls working.


> System error message?
> cat /var/adm/messages

No, my messages file is empty.


Any idea?

Thanks for your help.
Luca


--
Posted via Mailgate.ORG Server - http://www.Mailgate.ORG

In article <27975db4fba3be77913207dffa6e7285.60620@mygate.mai lgate.org>, "Luca Ferrari" <fluca1978@libero.it> writes:
> "Michael Tosch" <eedmit@NO.eed.SPAM.ericsson.PLS.se> wrote in message
> news:bv8ina$ch4$1@aken.eed.ericsson.se
>
>
> >
> > Things to try...
> >
> > Network problem?
> > ping -s server 1400 10
> > mkfile 10m 10m
> >
>
> The NFS/NIS server works well, is on line and export well username and
> password. Furthermore
> I can write a file on the mounted filesystem and I can see it on the
> file server
> (i.e., the write works) but cannot list. I can see also if I do 2echo
> file".

Especially with TCP transport it is a big difference between writing
a small file and a big file!

>
>
> > ls aliased to a bad command?
> > \ls -al
>
> No, there are not strange alias, just ls -F. Nevertheless even executing
> the
> full qualified command /bin/ls does not work.
>
>
> > PATH component on a dead NFS server?
> > /bin/ls -al
>
>
> No.
>
> > Which mount options?
> > mount | grep $PWD
> > df .
> >
>
>
> Here they are:
> bash-2.03$ mount | grep $PWD
> /home/luca on lx01:/export/SunOS-sun4u/home/luca
> remote/read/write/setuid/dev=3700028 on Wed Jan 28 16:56:09 2004
>
> bash-2.03$ df .
> /home/luca (lx01:/export/SunOS-sun4u/home/luca): 3559336 blocks
> 2862348 files
>
>
>
>
> > Why the shell cannot be changed?
> > /bin/sh
> > should start an sh (exit to return to previous shell)
> > exec /bin/sh
> > should replace current shell by sh
>
> Sorry, maybe I misdetailed the problem: I can of course change shell,
> but even in
> a different shell I cannot get the ls working.
>
>
> > System error message?
> > cat /var/adm/messages
>
> No, my messages file is empty.
>
>
> Any idea?
>

Well, many.
Some more questions:

What are the last system calls before ls hangs?
truss ls -la

Any bad patches or HW-related issues?
uname -a

Now let's concentrate on a corrupted installation.

Does one of the following work?
/usr/ucb/ls -la
/usr/xpg4/bin/ls -la

Does "showrev -p" look ok?

What yields
ldd /bin/ls

What yields
sum /bin/ls `ldd /bin/ls | awk '{print $NF}'`
and how does it compare with your working system?


--
Michael Tosch
IT Specialist
HP Managed Services Germany
Phone +49 2407 575 313

"Michael Tosch" <eedmit@NO.eed.SPAM.ericsson.PLS.se> wrote in message
news:bv8v97$2da$1@aken.eed.ericsson.se

>
> What are the last system calls before ls hangs?
> truss ls -la
>

Here's the output, but I don't know how to find bad things here:

~$ truss ls -la
execve("ls", 0xFFBEFCB4, 0xFFBEFCC0) argc = 2
open("/var/ld/ld.config", O_RDONLY) Err#2 ENOENT
open("/usr/lib/libc.so.1", O_RDONLY) = 3
fstat(3, 0xFFBEF384) = 0
mmap(0x00000000, 8192, PROT_READ|PROT_EXEC, MAP_PRIVATE, 3, 0) =
0xFF3A0000
mmap(0x00000000, 786432, PROT_READ|PROT_EXEC, MAP_PRIVATE, 3, 0) =
0xFF280000
mmap(0xFF338000, 24560, PROT_READ|PROT_WRITE|PROT_EXEC,
MAP_PRIVATE|MAP_FIXED, 3, 688128) = 0xFF338000
mmap(0xFF33E000, 6660, PROT_READ|PROT_WRITE|PROT_EXEC,
MAP_PRIVATE|MAP_FIXED|MAP_ANON, -1, 0) = 0xFF33E000
munmap(0xFF328000, 65536) = 0
mmap(0x00000000, 8192, PROT_READ|PROT_WRITE|PROT_EXEC,
MAP_PRIVATE|MAP_ANON, -1, 0) = 0xFF390000
memcntl(0xFF280000, 112228, MC_ADVISE, MADV_WILLNEED, 0, 0) = 0
close(3) = 0
open("/usr/lib/libdl.so.1", O_RDONLY) = 3
fstat(3, 0xFFBEF384) = 0
mmap(0xFF3A0000, 8192, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_FIXED, 3, 0)
= 0xFF3A0000
close(3) = 0
open("/usr/platform/SUNW,Ultra-5_10/lib/libc_psr.so.1", O_RDONLY) = 3
fstat(3, 0xFFBEF214) = 0
mmap(0x00000000, 8192, PROT_READ|PROT_EXEC, MAP_PRIVATE, 3, 0) =
0xFF380000
mmap(0x00000000, 16384, PROT_READ|PROT_EXEC, MAP_PRIVATE, 3, 0) =
0xFF370000
close(3) = 0
munmap(0xFF380000, 8192) = 0
dup(2) = 3
open("/dev/pts/01/uconf.inv", O_RDONLY) Err#13 EACCES
open("/dev/pts/01/uconf.inv", O_RDONLY) Err#13 EACCES
pathconf(".", _PC_PATH_MAX) = 1024
stat64("./", 0xFFBEF9D8) = 0
brk(0x00074988) = 0
brk(0x00076988) = 0
stat64("/", 0xFFBEF940) = 0
open64("./../", O_RDONLY|O_NDELAY) = 4
fcntl(4, F_SETFD, 0x00000001) = 0
fstat64(4, 0xFFBEEF60) = 0
fstat64(4, 0xFFBEF9D8) = 0
open("/etc/mnttab", O_RDONLY) = 5
fstat64(5, 0xFFBEE920) = 0
ioctl(5, TCGETA, 0xFFBEE8AC) Err#22 EINVAL
read(5, " / d e v / d s k / c 0 t".., 512) = 512
ioctl(5, (('m'<<8)|1), 0xFF33AC20) = 0
ioctl(5, (('m'<<8)|2), 0x00075BE0) = 0
read(5, " r e d\t / s h a r e d\t".., 512) = 147
lstat64("/home/luca", 0xFFBEEB40) = 0
lstat64("/home/luca/..", 0xFFBEEB40) = 0
llseek(5, 0, SEEK_CUR) = 659
close(5) = 0
close(4) = 0
fcntl(3, F_DUP2FD, 0x00000002) = 2
pipe() = 4 [5]
fork() = 4821
close(5) = 0
close(2) = 0
fstat64(4, 0xFFBEF908) = 0
brk(0x00076988) = 0
brk(0x00078988) = 0
ioctl(4, TCGETA, 0xFFBEF894) Err#22 EINVAL
read(4, 0x0007624C, 5120) = 0
llseek(4, 0, SEEK_CUR) Err#29 ESPIPE
close(4) = 0
llseek(0, 0, SEEK_CUR) = 8179
_exit(0)




> Any bad patches or HW-related issues?
> uname -a
>


~$ uname -a
SunOS -s 5.8 Generic_108528-05 sun4u sparc SUNW,Ultra-5_10

> Now let's concentrate on a corrupted installation.
>
> Does one of the following work?
> /usr/ucb/ls -la

It works rightly, but I don't know what's /usr/ucb. Since I'm not enough
expert with suns
(I have always worked mainly with linux) I don't know if this is good or
not. Can you explain
what's ucb is?

>
> What yields
> ldd /bin/ls

~$ ldd ls
ls:
libc.so.1 => /usr/lib/libc.so.1
libdl.so.1 => /usr/lib/libdl.so.1
/usr/platform/SUNW,Ultra-5_10/lib/libc_psr.so.1

>
> What yields
> sum /bin/ls `ldd /bin/ls | awk '{print $NF}'`
> and how does it compare with your working system?

~$ sum /bin/ls `ldd /bin/ls | awk '{print $NF}'`
58950 37 /bin/ls
7068 2207 /usr/lib/libc.so.1
26473 10 /usr/lib/libdl.so.1
41555 34 /usr/platform/SUNW,Ultra-5_10/lib/libc_psr.so.1

while on the working system is:
~$ sum /bin/ls `ldd /bin/ls | awk '{print $NF}'`
34850 37 /bin/ls
7068 2207 /usr/lib/libc.so.1
26473 10 /usr/lib/libdl.so.1
41555 34 /usr/platform/SUNW,Ultra-5_10/lib/libc_psr.so.1

Can you explain what does it mean?

Thanks for your help.

Luca


--
Posted via Mailgate.ORG Server - http://www.Mailgate.ORG

In article
<604905b75b661268523b275f34a8a42c.60620@mygate.mai lgate.org>,
"Luca Ferrari" <fluca1978@libero.it> wrote:

> pipe() = 4 [5]
> fork() = 4821
> close(5) = 0
> close(2) = 0
> fstat64(4, 0xFFBEF908) = 0
> brk(0x00076988) = 0
> brk(0x00078988) = 0
> ioctl(4, TCGETA, 0xFFBEF894) Err#22 EINVAL
> read(4, 0x0007624C, 5120) = 0
> llseek(4, 0, SEEK_CUR) Err#29 ESPIPE
> close(4) = 0

I think you've got a bogus "ls" command. I can't think of a reason why
it would need to fork a child process and talk to it with a pipe.

Maybe you've been rootkitted.

--
Barry Margolin, barmar@alum.mit.edu
Arlington, MA
*** PLEASE post questions in newsgroups, not directly to me ***

In article <604905b75b661268523b275f34a8a42c.60620@mygate.mai lgate.org>, "Luca Ferrari" <fluca1978@libero.it> writes:
> "Michael Tosch" <eedmit@NO.eed.SPAM.ericsson.PLS.se> wrote in message
> news:bv8v97$2da$1@aken.eed.ericsson.se
>
> >
> > What are the last system calls before ls hangs?
> > truss ls -la
> >
>
> Here's the output, but I don't know how to find bad things here:

it is wrong; it shouldnt pipe() and fork().

>
> ~$ truss ls -la
> execve("ls", 0xFFBEFCB4, 0xFFBEFCC0) argc = 2
> open("/var/ld/ld.config", O_RDONLY) Err#2 ENOENT
> open("/usr/lib/libc.so.1", O_RDONLY) = 3
> fstat(3, 0xFFBEF384) = 0
> mmap(0x00000000, 8192, PROT_READ|PROT_EXEC, MAP_PRIVATE, 3, 0) =
> 0xFF3A0000
> mmap(0x00000000, 786432, PROT_READ|PROT_EXEC, MAP_PRIVATE, 3, 0) =
> 0xFF280000
> mmap(0xFF338000, 24560, PROT_READ|PROT_WRITE|PROT_EXEC,
> MAP_PRIVATE|MAP_FIXED, 3, 688128) = 0xFF338000
> mmap(0xFF33E000, 6660, PROT_READ|PROT_WRITE|PROT_EXEC,
> MAP_PRIVATE|MAP_FIXED|MAP_ANON, -1, 0) = 0xFF33E000
> munmap(0xFF328000, 65536) = 0
> mmap(0x00000000, 8192, PROT_READ|PROT_WRITE|PROT_EXEC,
> MAP_PRIVATE|MAP_ANON, -1, 0) = 0xFF390000
> memcntl(0xFF280000, 112228, MC_ADVISE, MADV_WILLNEED, 0, 0) = 0
> close(3) = 0
> open("/usr/lib/libdl.so.1", O_RDONLY) = 3
> fstat(3, 0xFFBEF384) = 0
> mmap(0xFF3A0000, 8192, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_FIXED, 3, 0)
> = 0xFF3A0000
> close(3) = 0
> open("/usr/platform/SUNW,Ultra-5_10/lib/libc_psr.so.1", O_RDONLY) = 3
> fstat(3, 0xFFBEF214) = 0
> mmap(0x00000000, 8192, PROT_READ|PROT_EXEC, MAP_PRIVATE, 3, 0) =
> 0xFF380000
> mmap(0x00000000, 16384, PROT_READ|PROT_EXEC, MAP_PRIVATE, 3, 0) =
> 0xFF370000
> close(3) = 0
> munmap(0xFF380000, 8192) = 0
> dup(2) = 3
> open("/dev/pts/01/uconf.inv", O_RDONLY) Err#13 EACCES
> open("/dev/pts/01/uconf.inv", O_RDONLY) Err#13 EACCES
> pathconf(".", _PC_PATH_MAX) = 1024
> stat64("./", 0xFFBEF9D8) = 0
> brk(0x00074988) = 0
> brk(0x00076988) = 0
> stat64("/", 0xFFBEF940) = 0
> open64("./../", O_RDONLY|O_NDELAY) = 4
> fcntl(4, F_SETFD, 0x00000001) = 0
> fstat64(4, 0xFFBEEF60) = 0
> fstat64(4, 0xFFBEF9D8) = 0
> open("/etc/mnttab", O_RDONLY) = 5
> fstat64(5, 0xFFBEE920) = 0
> ioctl(5, TCGETA, 0xFFBEE8AC) Err#22 EINVAL
> read(5, " / d e v / d s k / c 0 t".., 512) = 512
> ioctl(5, (('m'<<8)|1), 0xFF33AC20) = 0
> ioctl(5, (('m'<<8)|2), 0x00075BE0) = 0
> read(5, " r e d\t / s h a r e d\t".., 512) = 147
> lstat64("/home/luca", 0xFFBEEB40) = 0
> lstat64("/home/luca/..", 0xFFBEEB40) = 0
> llseek(5, 0, SEEK_CUR) = 659
> close(5) = 0
> close(4) = 0
> fcntl(3, F_DUP2FD, 0x00000002) = 2
> pipe() = 4 [5]
> fork() = 4821
> close(5) = 0
> close(2) = 0
> fstat64(4, 0xFFBEF908) = 0
> brk(0x00076988) = 0
> brk(0x00078988) = 0
> ioctl(4, TCGETA, 0xFFBEF894) Err#22 EINVAL
> read(4, 0x0007624C, 5120) = 0
> llseek(4, 0, SEEK_CUR) Err#29 ESPIPE
> close(4) = 0
> llseek(0, 0, SEEK_CUR) = 8179
> _exit(0)
>
>
>
>
> > Any bad patches or HW-related issues?
> > uname -a
> >
>
>
> ~$ uname -a
> SunOS -s 5.8 Generic_108528-05 sun4u sparc SUNW,Ultra-5_10
>
> > Now let's concentrate on a corrupted installation.
> >
> > Does one of the following work?
> > /usr/ucb/ls -la
>
> It works rightly, but I don't know what's /usr/ucb. Since I'm not enough
> expert with suns
> (I have always worked mainly with linux) I don't know if this is good or
> not. Can you explain
> what's ucb is?

/usr/ucb should provide backward compatibility to SunOS 4.x.

>
> >
> > What yields
> > ldd /bin/ls
>
> ~$ ldd ls
> ls:
> libc.so.1 => /usr/lib/libc.so.1
> libdl.so.1 => /usr/lib/libdl.so.1
> /usr/platform/SUNW,Ultra-5_10/lib/libc_psr.so.1
>
> >
> > What yields
> > sum /bin/ls `ldd /bin/ls | awk '{print $NF}'`
> > and how does it compare with your working system?
>
> ~$ sum /bin/ls `ldd /bin/ls | awk '{print $NF}'`
> 58950 37 /bin/ls
> 7068 2207 /usr/lib/libc.so.1
> 26473 10 /usr/lib/libdl.so.1
> 41555 34 /usr/platform/SUNW,Ultra-5_10/lib/libc_psr.so.1
>
> while on the working system is:
> ~$ sum /bin/ls `ldd /bin/ls | awk '{print $NF}'`
> 34850 37 /bin/ls
> 7068 2207 /usr/lib/libc.so.1
> 26473 10 /usr/lib/libdl.so.1
> 41555 34 /usr/platform/SUNW,Ultra-5_10/lib/libc_psr.so.1
>
> Can you explain what does it mean?

Checksumming the binary. Also checksumming the used dynamic
link libraries.
And in fact the checksum of /bin/ls of your system is wrong.
It is corrupted.
You can copy it over from the working system.

I suspect a bad disk, that produces data corruptions.
Exchange it ASAP, and reinstall Solaris.
Data corruption can mean any sort of malfunction!

Pleas give your disk model by the command

iostat -E

or

format < /dev/null

Model ST38420A DateCode 99/2...99/8 is famous for
producing data corruptions.


--
Michael Tosch
IT Specialist
HP Managed Services Germany
Phone +49 2407 575 313

"Michael Tosch" <eedmit@NO.eed.SPAM.ericsson.PLS.se> wrote in message
news:bvajfu$n5r$1@aken.eed.ericsson.se

*> Pleas give your disk model by the command
>
> iostat -E
>

*~$ iostat -E
dad0 Soft Errors: 0 Hard Errors: 0 Transport Errors: 0
Model: ST34342A Revision: 0.42 Serial No: VGG3
Size: 4.30GB <4302789120 bytes>
Media Error: 0 Device Not Ready: 0 No Device: 0 Recoverable: 0
Illegal Request: 0



--
Posted via Mailgate.ORG Server - http://www.Mailgate.ORG

In article <f1a63f2b188b14662d8e63cdf91746b6.60620@mygate.mai lgate.org>, "Luca Ferrari" <fluca1978@libero.it> writes:
> "Michael Tosch" <eedmit@NO.eed.SPAM.ericsson.PLS.se> wrote in message
> news:bvajfu$n5r$1@aken.eed.ericsson.se
>
> *> Pleas give your disk model by the command
> >
> > iostat -E
> >
>
> *~$ iostat -E
> dad0 Soft Errors: 0 Hard Errors: 0 Transport Errors: 0
> Model: ST34342A Revision: 0.42 Serial No: VGG3
> Size: 4.30GB <4302789120 bytes>
> Media Error: 0 Device Not Ready: 0 No Device: 0 Recoverable: 0
> Illegal Request: 0
>
>

Ok, ST34342A is the 4GB variant of the famous 8GB disk.
I have seen hardly any data corruptions on them, but it is the
same (cheap/slow/bad) technology.
Exchange the disk!
If you get the same type on warranty, the data corruptions will
be gone, but you'll still have the low performance.
My recommendation: buy a good IDE disk elsewhere.

--
Michael Tosch
IT Specialist
HP Managed Services Germany
Phone +49 2407 575 313