Alex de Vaal
2008-Mar-21 13:02 UTC
[Samba] upgrade from 3.0.14a to 3.0.27a; CUPS log files flooding!
Hello, I've upgraded 2 of my Samba servers (out of 100+ Samba servers) in my live environment from version 3.0.14a to 3.0.27a. I'm using CUPS as print system for Samba (Samba is compiled with cups-devel) and I use the "Point 'n Print" mechanism of Samba (Windows) to provide XP users with the appropriate drives for the attached printers. CUPS is configured with RAW queues and all printers are connected via the HP JetDirectCard to the network. In CUPS I configured to send the print jobs to the JetDirectCard of the printer. So far so good, but after the upgrade from Samba 3.0.14a to 3.0.27a the logs files of CUPS (in /var/log/cups) are flooded, especially the access_log and error_log files. On a site with 100+ PC's and 20 CUPS printers these log files generate together 450 Mb of entries per day! With version 3.0.14a these log files generated 20 Mb of entries in one week? This happens on the Samba 3.0.27a packages of both RHEL and RHL9, so it is not the underlying Linux OS (RHEL4, CentOS4.x and RHL9) causing the problem. In the log files of CUPS I see entries like these flooding the error_log file: E [19/Mar/2008:13:13:30 +0100] get_printer_attrs: resource name '/printers' no good! E [19/Mar/2008:13:13:30 +0100] get_printer_attrs: resource name '/TEST_ADM06' no good! E [19/Mar/2008:13:13:30 +0100] get_printer_attrs: resource name '/TEST_ADM05' no good! E [19/Mar/2008:13:13:30 +0100] get_printer_attrs: resource name '/printers' no good! E [19/Mar/2008:13:13:30 +0100] get_printer_attrs: resource name '/printers' no good! E [19/Mar/2008:13:13:30 +0100] get_printer_attrs: resource name '/printers' no good! E [19/Mar/2008:13:13:31 +0100] get_printer_attrs: resource name '/printers' no good! E [19/Mar/2008:13:13:31 +0100] get_printer_attrs: resource name '/TEST_ADM06' no good! E [19/Mar/2008:13:13:31 +0100] get_printer_attrs: resource name '/TEST_ADM05' no good! E [19/Mar/2008:13:13:31 +0100] get_printer_attrs: resource name '/printers' no good! E [19/Mar/2008:13:13:31 +0100] get_printer_attrs: resource name '/TEST_ADM06' no good! E [19/Mar/2008:13:13:31 +0100] get_printer_attrs: resource name '/TEST_ADM05' In access_log I see entries like these flooding the file: localhost - - [16/Mar/2008:03:38:55 +0100] "POST / HTTP/1.1" 200 195 localhost - - [16/Mar/2008:03:38:55 +0100] "POST / HTTP/1.1" 200 278 localhost - - [16/Mar/2008:03:38:55 +0100] "POST / HTTP/1.1" 200 195 localhost - - [16/Mar/2008:03:38:55 +0100] "POST / HTTP/1.1" 200 199 localhost - - [16/Mar/2008:03:38:55 +0100] "POST / HTTP/1.1" 200 278 localhost - - [16/Mar/2008:03:38:55 +0100] "POST / HTTP/1.1" 200 199 localhost - - [16/Mar/2008:03:38:56 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:03:38:56 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:03:41:54 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:03:41:54 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:03:49:56 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:03:49:56 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:03:51:41 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:03:51:41 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:03:53:04 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:03:53:04 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:04:00:56 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:04:00:56 +0100] "POST / HTTP/1.1" 200 134 Printing from XP clients to the mentioned printers works fine, also when I create a new printer in CUPS it shows up in Samba and I can upload the windows printers drivers to the Samba server.>From an XP client I can connect to this Samba printer and send a print job,which is being printed by the printer via CUPS! To avoid flooding of the CUPS log files I've changed the following in /etc/cups/cupsd.conf: ========================================= #AccessLog /var/log/cups/access_log AccessLog /dev/null # info Log all requests and state changes. (default) # warn Log errors and warnings. # error Log only errors. # none Log nothing. # LogLevel none ========================================= This result in 0 bytes log files and no real logging of CUPS, besides the page_log file. Can anyone tell me if they encountered the same issue and what they did to solve it? My smb.conf looks like this: # Global parameters [global] workgroup = TEST realm = TEST.COM server string = %h server (Samba %v) security = ADS password server = adm04.test.com, adm01.test.com log file = /var/log/samba/%m.log max log size = 200 socket options = TCP_NODELAY SO_RCVBUF=8192 SO_SNDBUF=8192 printcap cache time = 660 domain master = No ldap timeout = 15 ldap page size = 100 idmap uid = 10000-20000 idmap gid = 10000-20000 template homedir = /data/hom/%U template shell = /bin/bash winbind cache time = 660 printer admin = root, "@TEST.COM\Domain Admins", @TEST.COM\DEP_TEST_IT oplocks = No level2 oplocks = No #Below Global Settings Needed For Samba 3.0.27a default devmode = No enable privileges = Yes host msdfs = No msdfs root = No winbind enum users = Yes winbind enum groups = Yes winbind nested groups = No printing = cups strict locking = Yes [printers] comment = All Printers path = /var/spool/samba guest ok = Yes printable = Yes browseable = No [print$] comment = Printer Driver Download Area path = /etc/samba/drivers write list = root, "@TEST.COM\Domain Admins", @TEST.COM\DEP_TEST_IT force user = root guest ok = Yes Regards, Alex.
adrian sender
2008-Mar-24 12:46 UTC
[Samba] upgrade from 3.0.14a to 3.0.27a; CUPS log files flooding!
I have observed a similar problem; my main concern was cpu load caused by samba querying cups for printer "comment" and "description" fields instead of its local database. Set this through windows by mapping the printers as a print admin user. Once fields were set on all printers and samba restarted cpu load dropped from 50+% to almost 0. This should also resolve logging problems. Regards, Adrian Sender --Forwarded Message Attachment--From: samba.alex@gmail.com Subject: [Samba] upgrade from 3.0.14a to 3.0.27a; CUPS log files flooding! Date: Fri, 21 Mar 2008 14:02:23 +0100 To: samba@lists.samba.org Hello, I've upgraded 2 of my Samba servers (out of 100+ Samba servers) in my live environment from version 3.0.14a to 3.0.27a. I'm using CUPS as print system for Samba (Samba is compiled with cups-devel) and I use the "Point 'n Print" mechanism of Samba (Windows) to provide XP users with the appropriate drives for the attached printers. CUPS is configured with RAW queues and all printers are connected via the HP JetDirectCard to the network. In CUPS I configured to send the print jobs to the JetDirectCard of the printer. So far so good, but after the upgrade from Samba 3.0.14a to 3.0.27a the logs files of CUPS (in /var/log/cups) are flooded, especially the access_log and error_log files. On a site with 100+ PC's and 20 CUPS printers these log files generate together 450 Mb of entries per day! With version 3.0.14a these log files generated 20 Mb of entries in one week? This happens on the Samba 3.0.27a packages of both RHEL and RHL9, so it is not the underlying Linux OS (RHEL4, CentOS4.x and RHL9) causing the problem. In the log files of CUPS I see entries like these flooding the error_log file: E [19/Mar/2008:13:13:30 +0100] get_printer_attrs: resource name '/printers' no good! E [19/Mar/2008:13:13:30 +0100] get_printer_attrs: resource name '/TEST_ADM06' no good! E [19/Mar/2008:13:13:30 +0100] get_printer_attrs: resource name '/TEST_ADM05' no good! E [19/Mar/2008:13:13:30 +0100] get_printer_attrs: resource name '/printers' no good! E [19/Mar/2008:13:13:30 +0100] get_printer_attrs: resource name '/printers' no good! E [19/Mar/2008:13:13:30 +0100] get_printer_attrs: resource name '/printers' no good! E [19/Mar/2008:13:13:31 +0100] get_printer_attrs: resource name '/printers' no good! E [19/Mar/2008:13:13:31 +0100] get_printer_attrs: resource name '/TEST_ADM06' no good! E [19/Mar/2008:13:13:31 +0100] get_printer_attrs: resource name '/TEST_ADM05' no good! E [19/Mar/2008:13:13:31 +0100] get_printer_attrs: resource name '/printers' no good! E [19/Mar/2008:13:13:31 +0100] get_printer_attrs: resource name '/TEST_ADM06' no good! E [19/Mar/2008:13:13:31 +0100] get_printer_attrs: resource name '/TEST_ADM05' In access_log I see entries like these flooding the file: localhost - - [16/Mar/2008:03:38:55 +0100] "POST / HTTP/1.1" 200 195 localhost - - [16/Mar/2008:03:38:55 +0100] "POST / HTTP/1.1" 200 278 localhost - - [16/Mar/2008:03:38:55 +0100] "POST / HTTP/1.1" 200 195 localhost - - [16/Mar/2008:03:38:55 +0100] "POST / HTTP/1.1" 200 199 localhost - - [16/Mar/2008:03:38:55 +0100] "POST / HTTP/1.1" 200 278 localhost - - [16/Mar/2008:03:38:55 +0100] "POST / HTTP/1.1" 200 199 localhost - - [16/Mar/2008:03:38:56 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:03:38:56 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:03:41:54 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:03:41:54 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:03:49:56 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:03:49:56 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:03:51:41 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:03:51:41 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:03:53:04 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:03:53:04 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:04:00:56 +0100] "POST / HTTP/1.1" 200 134 localhost - - [16/Mar/2008:04:00:56 +0100] "POST / HTTP/1.1" 200 134 Printing from XP clients to the mentioned printers works fine, also when I create a new printer in CUPS it shows up in Samba and I can upload the windows printers drivers to the Samba server.>From an XP client I can connect to this Samba printer and send a print job,which is being printed by the printer via CUPS! To avoid flooding of the CUPS log files I've changed the following in /etc/cups/cupsd.conf: ========================================= #AccessLog /var/log/cups/access_log AccessLog /dev/null # info Log all requests and state changes. (default) # warn Log errors and warnings. # error Log only errors. # none Log nothing. # LogLevel none ========================================= This result in 0 bytes log files and no real logging of CUPS, besides the page_log file. Can anyone tell me if they encountered the same issue and what they did to solve it? My smb.conf looks like this: # Global parameters [global] workgroup = TEST realm = TEST.COM server string = %h server (Samba %v) security = ADS password server = adm04.test.com, adm01.test.com log file = /var/log/samba/%m.log max log size = 200 socket options = TCP_NODELAY SO_RCVBUF=8192 SO_SNDBUF=8192 printcap cache time = 660 domain master = No ldap timeout = 15 ldap page size = 100 idmap uid = 10000-20000 idmap gid = 10000-20000 template homedir = /data/hom/%U template shell = /bin/bash winbind cache time = 660 printer admin = root, "@TEST.COM\Domain Admins", @TEST.COM\DEP_TEST_IT oplocks = No level2 oplocks = No #Below Global Settings Needed For Samba 3.0.27a default devmode = No enable privileges = Yes host msdfs = No msdfs root = No winbind enum users = Yes winbind enum groups = Yes winbind nested groups = No printing = cups strict locking = Yes [printers] comment = All Printers path = /var/spool/samba guest ok = Yes printable = Yes browseable = No [print$] comment = Printer Driver Download Area path = /etc/samba/drivers write list = root, "@TEST.COM\Domain Admins", @TEST.COM\DEP_TEST_IT force user = root guest ok = Yes Regards, Alex. _________________________________________________________________ Fashion, beauty, health, relationship advice and horoscopes. http://ninemsn.com.au/share/redir/adTrack.asp?mode=click&clientID=873&referral=MarchHotmailTagline&URL=http://lifestyle.ninemsn.com.au/?ocid=T003HOT40A0710G
Alex de Vaal
2008-Mar-27 11:51 UTC
[Samba] upgrade from 3.0.14a to 3.0.27a; CUPS log files flooding!
On Mon Mar 24 at 12:39:30, Adrian Sender wrote:> I have observed a similar problem; my main concern was cpu load caused bysamba querying cups> for printer "comment" and "description" fields instead of its localdatabase.> Set this through windows by mapping the printers as a print admin user.> Once fields were set on all printers and samba restarted cpu load droppedfrom 50+% to almost 0.> This should also resolve logging problems.My users get the Windows (Samba/CUPS) printers installed via the logon script of the W2k3 AD domain. The windows users have user level on the XP client. If I add those users (as member of a AD group) to the "printer admin =" in smb.conf, then the problem should vanish? Thnx for your reply, Alex.