samba - Sep 2005 - Solaris NSS/winbind problem - large groups

Hello All,

It seems that 'getent group groupname' hangs on large groups for Solaris
8/9
running winbind. Currently running Samba 3.0.20 but also tried 3.0.10. 
'ls -l' also hangs if listing a file created by a user who's primary
group
is too large(such as Domain Users).  Users can still access the files but
netbackups hang.

The problem I am experiencing is found in the archives here:
http://marc.theaimsgroup.com/?l=samba-technical&m=107840151432362&w=2  ,
but
I don't see a resolution.  Is there a workaround or patch for this?

I originally contacted Sun about this problem and also another issue that I
thought was similar, relating to large ACL's.  They agreed to patch
ACLtotext but not the NSS.
Sun just released a patch for the ACLtotext bug (120441-01 Sol9 sparc) for
large ACL's, if you are interested.  


Thanks,

Kevin

Does anyone know if this is a problem specific to Solaris or if there is an
existing fix or one on the way? Any help is greatly appreciated.

'getent group "Domain Users"' (or any other large group) 
-hangs
'getent group' 	-does not return a full group listing.
'wbinfo -g' 	-lists all groups

partial truss output of getent group "Domain Users" (Repeats):

alarm(0)                                        = 0
sigaction(SIGALRM, 0xFFBFD6F8, 0xFFBFD7A8)      = 0
sigfillset(0xFF242AA8)                          = 0
sigprocmask(SIG_BLOCK, 0xFFBFD798, 0xFFBFD788)  = 0
alarm(1)                                        = 0
    Received signal #14, SIGALRM, in sigsuspend() [caught]
sigsuspend(0xFFBFD778)                          Err#4 EINTR
setcontext(0xFFBFD460)
alarm(0)                                        = 0
sigprocmask(SIG_UNBLOCK, 0xFFBFD798, 0x00000000) = 0
sigaction(SIGALRM, 0xFFBFD6F8, 0x00000000)      = 0
alarm(0)                                        = 0
sigaction(SIGALRM, 0xFFBFD6F8, 0xFFBFD7A8)      = 0
sigprocmask(SIG_BLOCK, 0xFFBFD798, 0xFFBFD788)  = 0
alarm(2)                                        = 0
sigsuspend(0xFFBFD778)          (sleeping...)
    Received signal #14, SIGALRM, in sigsuspend() [caught]
sigsuspend(0xFFBFD778)                          Err#4 EINTR
setcontext(0xFFBFD460)
alarm(0)                                        = 0
sigprocmask(SIG_UNBLOCK, 0xFFBFD798, 0x00000000) = 0
sigaction(SIGALRM, 0xFFBFD6F8, 0x00000000)      = 0
alarm(0)                                        = 0
sigaction(SIGALRM, 0xFFBFD6F8, 0xFFBFD7A8)      = 0
sigprocmask(SIG_BLOCK, 0xFFBFD798, 0xFFBFD788)  = 0
alarm(4)                                        = 0
sigsuspend(0xFFBFD778)          (sleeping...)

In a previous list in samba-technical from last year, Andrew Bartlett wrote:
"From what I understand, the issue is not in winbindd, or wbinfo - but in
nss_winbindd. So the command to run under truss is 'id' or 'getent
group'
etc. This is why I think this is a solaris specific sun (or
winbind_nss_solaris.c) bug. Andrew Bartlett"


Thanks- ,

Kevin Hutching