At some point, we need to implement OTP support for the HTML5 client.
Fixed in: r27076 r27077 r27078 Time reporting on bug 4132. The tester should check: * Authentication with normal passwords, as well as extra prompts ie OTP * Cleanup of stale "pamtester" processes and FIFOs in /tmp * Bonus: Authentication with passwordless username * Security aspects * Error handling
(In reply to comment #1) > * Authentication with normal passwords, as well as extra prompts ie OTP Work as expected, tested using pam_prompt.so. > * Bonus: Authentication with passwordless username Works in a strange way, empty password from login form is passed as response to OTP prompt, however the thinlinc client works this way too.
> * Error handling I configured pam_radius module for sshd which pointed out a non existing radius server, pamtester auth failures log auth failures into the html page but no trace in /var/log/tlwebaccess.log
While using pam_prompt and at login form when prompt for second input I stopped pamtester 'kill -STOP <pid>'. Then i continued with the login form filling in a prompt and continue, the form hangs forever (>60 secs). Maybe we should have a timeout for IPC tlwebaccess <-> pamtester
> * Cleanup of stale "pamtester" processes and FIFOs in /tmp After the stop of pamtester process in comment #4 it took some time for cleanup but /tmp was cleaned up from .err/.in/.out fifos, however the tlwebaccess process which wants to communicate with pamtester seems to never die.
(In reply to comment #3) > > * Error handling > > I configured pam_radius module for sshd which pointed out a non existing radius > server, pamtester auth failures log auth failures into the html page but no > trace in /var/log/tlwebaccess.log Moved to bug https://www.cendio.com/bugzilla/show_bug.cgi?id=4632.
(In reply to comment #5) > > * Cleanup of stale "pamtester" processes and FIFOs in /tmp > > After the stop of pamtester process in comment #4 it took some time for cleanup > but /tmp was cleaned up from .err/.in/.out fifos, however the tlwebaccess > process which wants to communicate with pamtester seems to never die. This and comment #4 should be fixed in 27303.
(In reply to comment #7) > (In reply to comment #5) > > > * Cleanup of stale "pamtester" processes and FIFOs in /tmp > > > > After the stop of pamtester process in comment #4 it took some time for cleanup > > but /tmp was cleaned up from .err/.in/.out fifos, however the tlwebaccess > > process which wants to communicate with pamtester seems to never die. > > This and comment #4 should be fixed in 27303. Using build 3937, Verified that there is a 120 seconds timeout for IPC communications tlwebaccess <-> pamtester. Upon timeout, the rendered html page shows an error message.
Also tested to login steps until the OTP prompt, were i went back to main login form (a few times), leaving a few tlwebaccess+pamtester alive but not used. Those left overs where successfully removed after a timeout. After these tests and fixes it seems to work fine. Closing this bug as fixed.