Hi all; We have Squid 2.5.STABLE7running with 30 ntlm_auth helpers Version 3.0.10-1.fc3. The problem is that on the squid?s cachemgr.cgi->NTLM User Authenticator Stats is possible to verify that the ntlm_auth processes are slowly having the flag R (Reserved or Deferred) set and never being used again (the number of requests stops). This problem goes until there is no more ntlm_auth process available and the squid restarts itself, restarting all ntlm_auth too. After the restart, everything is goes back to normal and the problem repeats. We detected that the R flag appears more agressively if the domain controller is under more load (like running a backup script). Another strange behavior is that the TIME on the cachemgr.cgi->NTLM User Authenticator Stats for some ntlm_auth keeps growing with no flag set at all. Well, the questions are: 1- What does each of the flags mean? They are: A = ALIVE B = BUSY C = CLOSING R = RESERVED or DEFERRED S = SHUTDOWN P = PLACEHOLDER 2- Any ideas why the R flag is spreading throught all the ntlm_auth processes? 3- Why the time for some of the ntlm_auth processes keeps growing with no authentication to do (I think...), like processes 11 and 12 below ? # FD PID # Requests Flags Time Offset Request 1 8 8656 19005 R 0.155 0 (none) 2 9 8657 10502 R 8.135 0 (none) 3 10 8658 4355 0.135 0 (none) 4 11 8659 1547 0.561 0 (none) 5 12 8660 602 0.561 0 (none) 6 13 8661 230 21.210 0 (none) 7 14 8662 107 21.210 0 (none) 8 15 8663 64 21.210 0 (none) 9 16 8664 28 50.800 0 (none) 10 17 8665 18 202.638 0 (none) 11 18 8666 8 202.605 0 (none) 12 19 8667 2 219.373 0 (none) 13 20 8668 0 0.000 0 (none) 14 21 8669 0 0.000 0 (none) 15 22 8670 0 0.000 0 (none) 16 23 8671 0 0.000 0 (none) 17 24 8672 0 0.000 0 (none) 18 25 8673 0 0.000 0 (none) 19 26 8674 0 0.000 0 (none) 20 27 8675 0 0.000 0 (none) 21 28 8676 0 0.000 0 (none) 22 29 8677 0 0.000 0 (none) 23 30 8678 0 0.000 0 (none) 24 31 8679 0 0.000 0 (none) 25 32 8680 0 0.000 0 (none) 26 33 8681 0 0.000 0 (none) 27 34 8682 0 0.000 0 (none) 28 35 8683 0 0.000 0 (none) 29 36 8684 0 0.000 0 (none) 30 37 8685 0 0.000 0 (none) Rafael Sarres de Almeida Se??o de Gerenciamento de Rede Superior Tribunal de Justi?a Tel: (61) 319-9342
On Fri, 2005-06-17 at 11:38 -0300, Rafael.Almeida@stj.gov.br wrote:> Hi all; > We have Squid 2.5.STABLE7running with 30 ntlm_auth helpers Version > 3.0.10-1.fc3. The problem is that on the squid?s cachemgr.cgi->NTLM User > Authenticator Stats is possible to verify that the ntlm_auth processes > are slowly having the flag R (Reserved or Deferred) set and never being > used again (the number of requests stops). This problem goes until there > is no more ntlm_auth process available and the squid restarts itself, > restarting all ntlm_auth too. After the restart, everything is goes back > to normal and the problem repeats. We detected that the R flag appears > more agressively if the domain controller is under more load (like running > a backup script).Perhaps this question is best directed at the squid list? While I wrote ntlm_auth, I'm not sure I can help with how squid is marking the state of the helpers. It might be that we need to have a better way to have ntlm_auth tell Squid that there is a problem now, but it might go away (previous helper designs had to be restarted for that to happen, but ntlm_auth can recover on it's own). Andrew Bartlett -- Andrew Bartlett http://samba.org/~abartlet/ Samba Developer, SuSE Labs, Novell Inc. http://suse.de Authentication Developer, Samba Team http://samba.org Student Network Administrator, Hawker College http://hawkerc.net -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 189 bytes Desc: This is a digitally signed message part Url : http://lists.samba.org/archive/samba/attachments/20050621/2dce105e/attachment.bin