* Grzegorz Pawel Szostak said: > > > > Playing BOFH is not always justified :)) > > > You are right, my question is: > > > If in /etc/passwd file user has in shell field an bash shell, and if s/he > > > after login will change his/her shell environment (limits) that is in bash > > > will be in ksh, etc.. ? > > If you're using lshell, yes. If you're using ulimit, no. If you use PAM, > > yes. > So why lshell + PAM doesn't work ? Either this or that, both is unnecessary. I use lshell because I'm to lazy to switch entirely to PAM yet :)))) > > > And all programs (ssh, pop, ...) will work ? > > Yes. They all use getpwent or derivatives and they in turn use the NSS > > libaries - that's, of course, if you use glibc. So they will work > > flawlessly. > It looks good :) And works good :)) > > > > What for? kernel implements everything you need (man setrlimit), you just > > > > need a program to set the limits for you. Test the debian/potato PAM - it > > > > does it allright. > > > I've tested PAM with lshell and it wasn't working ... > > Hmm... it works for me just fine. > Are you sure: PAM + lshell (i'm using potato distribution ) yes. The only thing that didn't work with PAM pam_limits module was the exclusion rule (the '-' line) - I sent the patch to the maintainer and it already works ok. lshell always worked it's just that its restrictions are not that fine-grained as those of pam_limits. marek
Attachment:
pgpnriokMbtO1.pgp
Description: PGP signature