+++ This bug was initially created as a clone of Bug #632568 +++ Description of problem: I added following line in /etc/security/limits.conf file - mi2 hard maxlogins 2 and logged as mi2 in tty1 and tty2. while trying to login to tty3 as mi2 it said "Too many logins" respawned the login process. Now i logged out from tty1 and tty2...I can't login as mi2 I can verify this using who command. Version-Release number of selected component (if applicable): How reproducible: 1.set maxlogin to 2 for user1 2.user1 logs into two different tty and another try gave message as "Too many login" 3.Now logout from early logged machine. 4.Now try and login as user1 again. you will error message "Too many users" Steps to Reproduce: 1.set maxlogin to 2 for user1 2.user1 logs into two different tty and another try gave message as "Too many login" 3.Now logout from early logged machine. 4.Now try and login as user1 again. you will error message "Too many users" Actual results: Unable to login Expected results: user must be allowed to login Additional info: Using Fedora 12. --- Additional comment from tmraz on 2010-09-13 10:55:27 CEST --- Can you please post the rpm -q upstart output on your machine? --- Additional comment from lakshmipathi.g on 2010-09-13 11:27:20 CEST --- Here its $ rpm -q upstart upstart-0.3.11-3.fc12.x86_64 $uname -a Linux space 2.6.32.16-141.fc12.x86_64 #1 SMP Wed Jul 7 04:49:59 UTC 2010 x86_64 x86_64 x86_64 GNU/Linux --- Additional comment from tmraz on 2010-09-13 11:41:25 CEST --- Please update upstart to 0.3.11-4.fc12 - this is in Fedora 12 updates repository. --- Additional comment from lakshmipathi.g on 2010-09-13 12:00:38 CEST --- upgraded to current version. # rpm -q upstart upstart-0.3.11-4.fc12.x86_64 Testing with user name : ll #grep 'maxlogins' /etc/security/limits.conf ll hard maxlogins 2 # >/var/log/secure who shows only laks: # who laks tty1 2010-09-13 15:19 (:0) laks pts/0 2010-09-13 15:20 (:0.0) I tried to login as ll ,as described above steps - # cat /var/log/secure Sep 13 15:25:09 localhost login: pam_limits(login:session): Too many logins (max 2) for ll Sep 13 15:25:09 localhost login: pam_unix(login:session): session opened for user ll by ll(uid=0) Sep 13 15:25:10 localhost login: pam_limits(login:session): Too many logins (max 2) for ll Sep 13 15:25:10 localhost login: Permission denied Is this a know issue/bug ? It's not yet resolved. --- Additional comment from lakshmipathi.g on 2010-09-13 12:02:18 CEST --- Please let me know,If you need more info. --- Additional comment from tmraz on 2010-09-13 12:22:48 CEST --- Have you rebooted the computer after the upgrade? --- Additional comment from lakshmipathi.g on 2010-09-13 12:52:48 CEST --- Yes,after yum install , i rebooted the machine and tested with user "ll". I have Selinux enabled. --- Additional comment from lakshmipathi.g on 2010-09-13 13:12:15 CEST --- Strange, it seems like working now for the first time.-I can live with that -,but not everytime. ========= see who shows only two logins - [root@space laks]# who new1 tty5 2010-09-13 16:26 new1 tty3 2010-09-13 16:27 while trying to login it shows - # cat /var/log/secure Sep 13 16:36:20 localhost login: pam_limits(login:session): Too many logins (max 3) for new1 Sep 13 16:36:20 localhost login: pam_unix(login:session): session opened for user new1 by new1(uid=0) Sep 13 16:36:20 localhost login: pam_limits(login:session): Too many logins (max 3) for new1 Sep 13 16:36:20 localhost login: Permission denied =========== (above log not related to below steps) -------- Steps to follow - to reproduce: 0.set maxlogin limit as 3 1.login as new1 in tty1,tty2,tty3 2.whiling logging to tty4 it says "Too many logins" 3.logout of tty3 4.login to tty4 it work now. But If i randomly logout and login again. this issue comes again. 5.logout of 1,2 6.login to tty5 it says "too many logins" Something is wrong,i don't know what it is. --- Additional comment from lakshmipathi.g on 2010-09-14 08:47:17 CEST --- It works if i wait for a minute and then login again.the issue is fixed for me. --- Additional comment from tmraz on 2010-09-14 09:08:33 CEST --- It seems like upstart does not clean the USER_PROCESS entries immediately.
Why is this cloned against systemd?
# ls -l /sbin/init lrwxrwxrwx. 1 root root 14 Sep 21 12:47 /sbin/init -> ../bin/systemd # rpm -q systemd systemd-10-3.fc15.x86_64 # cat /etc/security/limits.conf test hard maxlogins 2 login and logout user test on tty1, login and logout user test on tty2, login user test on tty3: Sep 22 08:51:04 rawhide-systemd login: pam_limits(login:session): Too many logins (max 2) for test Sep 22 08:51:04 rawhide-systemd login: pam_unix(login:session): session opened for user test by LOGIN(uid=0) Sep 22 08:51:04 rawhide-systemd login: Permission denied # w 08:52:33 up 5 min, 3 users, load average: 0.08, 0.08, 0.04 USER TTY FROM LOGIN@ IDLE JCPU PCPU WHAT root ttyS0 - 08:49 0.00s 0.14s 0.05s w
Created attachment 448863 [details] utmp dump after too many logins
Tools like pam, w, destop switch user, ... use utmp table to discover information who is currently logged, see utmp(5). systemd should at least set DEAD_PROCESS for utmp entries with already dead tty processes to avoid this issue.
Fixed now upstream.