After one failed authentication, winbind refuses all next authentication attempts with NT_STATUS_INVALID_HANDLE message. The same with 3.0.14a was without problems. Problem is reproduceable anytimes, for example with wbinfo -a command. winbindd_v1.patch only gives more info, the problem outlasts. Expected results are obtained form wbinfo -t, wbinfo -u, etc. anytime, only wbinfo -m gives nothing (already before failed auth). PDC is samba-3.0.14a. I tried some smb.conf changes (caching, schannel, signing, spnego), but without any effect.
Created attachment 1443 [details] Steps to reproduce Steps to reproduce with winbind responses
Created attachment 1444 [details] winbind debug winbindd -Find 10 output form wbinfo attempts from previous attachement
Just a quick info: Got it reproduced. The credential chain check fails on the 3.0.14a PDC. Strange.... Volker
Created attachment 1446 [details] Possible fix Could you try the attached patch? This fixes it for me. This is *very* weird, I tested the original and this version against NT4, W2k3 and Samba3.0.14a. Only Samba3.0.14a complained, the others seemed a lot more liberal. Jerry, Jeremy: Could you take a look at the patch? Thanks, Volker
Volker can you pull jcmd and maybe jpjanoski into this as I believe this may be related to a change they made here (from svn blame on 3.0) : 9112 jmcd /* moved from right after deal_with_creds above, since we weren't 9112 jmcd supposed to update unless logon was successful */ 9112 jmcd 9112 jmcd reseed_client_creds(&p->dc.clnt_cred, &q_u->sam_id.client.cred); 9112 jmcd memcpy(&p->dc.srv_cred, &p->dc.clnt_cred, sizeof(p->dc.clnt_cred)); I've changed this for HEAD and I think I need to check this more carefully to ensure I've got this right. Jeremy.
Created attachment 1447 [details] Patch for HEAD. Here's an attachment showing how I've fixed this in HEAD. I think this is correct (although it's hard to check right now as I'm out of the office). Jeremy.
Attached patch (1446) fixed problem for me too. Thanks. Only problem with "wbinfo -m" (no output) remains (only cosmetic problem :-). Do you have this problem too, or it's only some my misconfiguration?
Created attachment 1451 [details] Patch I've applied to 3.0 Ok, here's how I've fixed this in SVN. Jeremy.
Fixed by svn patch I think. Jeremy.