ANY HOPE GETTING RESPONSE TO QUESTIONS? - samba3.0alpha15 - s olar is 8

Thomas, Daniel J. Daniel.Thomas at
Wed Feb 27 08:10:16 GMT 2002

Just replying to let you know I'm listening =)...
Sorry Dave, I'm running 2.2.3a on my Solaris 8 machine and even that still
has some issues with winbind.

By the way, while I have the attention of you Solaris Gurus...  I was trying
to make my Ultra 1 boot from an external disk on the same scssi channel as
the default boot disk.  I changed my /etc/vfstab for the new disk to have
the right device references.  Then I went to prom level and used "setenv
boot-device disk2" (c0t2d0s0).  It boots fine, but after I repartitioned
disk0, I get some kind of mounting error when I rebooted like it's still
trying to mount "/" as c0t0d0s0.  Before I reconfigured disk0 it booted
without an error and a "df -k" didn't show c0t0d0s0 mounted and correctly
returned c0t2d0s0 mounted as "/".  After re-partitioning disk0 with
partition 0 as the free hog, I wasn't able to create the file system on it.
It insisted that it couldn't make the file system because "/dev/dsk/c0t2d0s0
was mounted as /"  At one point I noticed that the link that sun uses to
link the device files in /dev/dsk to the actual devices had c0t0d0s0 mapped
to scssi ID 2 rather then scssi ID 0.  I went to prom and did a "boot -r"
and this fixed the problem, but I still couldn't "newfs /dev/rdsk/c0t0d0s0".

Is this behavior some sort of OS bug?  Does Solaris make a link from
c0t0d0s0 to your boot device for some kind of compatibility issue?


-----Original Message-----
From: David Edward Shapiro [mailto:David.Edward.Shapiro at]
Sent: Wednesday, February 27, 2002 9:08 AM
To: 'samba at'
Cc: 'samba-technical at'
Subject: ANY HOPE GETTING RESPONSE TO QUESTIONS? - samba3.0alpha15 -
solar is 8

Several questions posted with no reponse for solaris 8, samba3.0alpha15:

*	testparm whining about valid.dat, users.dat, etc.  (If you copy
those files from the source directory and plop them in /usr/local/samba/lib,
my users can not map a drive or send files with smbclient to nt share.)
Testparm output errors:

	incorrect size for /usr/local/samba/lib/valid.dat - got 0 expected
	creating default valid table
	incorrect size for /usr/local/samba/lib/valid.dat - got 0 expected
	creating default valid table

*	How do you stop this handle_source_env error:
	handle_source_env: Failed to open file , Error was No such file or
	Unable to copy service - source not found: 

*	Why does testparm complain about can't find include file?
	Can't find include file 

*	Why does samba after recent changes to alpha15 now complain about
conversion not supported?
	Conversion from UCS-2LE to UTF-8 not supported
	Conversion from ASCII to UTF-8 not supported
	Conversion from ASCII to UTF-8 not supported
	Conversion from ASCII to UTF-8 not supported
	Conversion from UTF-8 to UCS-2LE not supported
	Conversion from UTF-8 to ASCII not supported
	Conversion from UTF-8 to ASCII not supported
	Conversion from UTF-8 to ASCII not supported

*	Why does getent passwd on solaris return some junk in output?
	INS+DavidSha:x,ha:10304:1000:David Edward
	Note the x,ha

*	Why if wbinfo -u/-g/-t returns correct results, getent passwd/getent
group returns lines (although Domain Users group does not display because it
is too big for solaris version and user line has junk as shown in previous
item?!?) doesn't samba let these users gain access to my shares if I list
them as valid users (e.g., valid users = INS+DavidSha).  The log for the
workstation trying to connect says invalid user davidsha instead of checking
for INS+DavidSha:

	check_password:  Authenticaion for user [davidsha] -> [davidsha]
	Note spelling error for Authentication

	What needs to change to get it to look for INS+DavidSha instead of
davidsha?  The only way I can get this working is to create a user called
davidsha in /etc/passwd, which circumvents the use of winbindd.  I have
winbindd working on linux fine and another solaris 8 box.  The difference is
that this box is running in 32 bit mode, and the other is running in 64 bit
mode.  I downloaded gcc 3.0.4, the lastest binutils, libtools and buildt
them fresh yesterday and then rebuildt samba, but it did not fix any of the
above issues.  Note that the other solaris box has the garbage in the getent
passwd too, but it authenticates INS+DavidSha, which is why it is working.

	Note: If I add the user davidsha in /etc/passwd (i.e.,
davidsha:x:1019:1000:NT User:/dev/null:/bin/false) and then run smbclient -L
enterprise -U INS+DavidSha, I get a list of shares?!?  If I remove the
davidsha entry the smbclient command fails
*	wbinfo -t gets secret good and every few tries it says bad.  Why
would it be losing the secret and getting bad like this?  
	Note that for the previous bullet, I can run the smbclient over and
over and over and it will never work if entry is not in /etc/passwd, but it
will work with entry in /etc/passwd in same sporadic way as the secret fails
and recovers.

David E. Shapiro
Senior Unix Admin <<log.smbd.txt>>  <<log.winbindd.txt>>  <<log.nmbd.txt>>
<<log.bti-c2433.txt>>  <<smb.conf>> BTi - the future of communications4300
Six Forks Road, Raleigh, NC 27609

More information about the samba-technical mailing list