<P>Any interventions at the file level within Mercury, except policies invoked, is considered dangerous. That said, we manipulate several config-files automatically without problems, but never queue directories - and it has worked without problems for a decade and a half now.</P>

<p>We recently upgraded to Mercury 4.72 and Windows 2008 R2 on 32 bit architecture.</p><p>Occasionally , under high volumes our Mercury mail Queue is failing to process correctly. The Mercury service seems to be CPU bound, consuming 12%CPU on a Quad CPU for hours. No email is delivered. A restart of  the service un-hangs it and the queue is quickly processed. </p><p>The log files show no errors.</p><p> </p><p>We operate Mercury mostly as an internal  SMTP relay engine to push PDF docs through a larger Exchange PostOffice.</p><p>  </p><p> Below is our INI file. Any suggestions are greatly appreciated.</p><p>Peter Dunphy</p><p> </p><p> #  MERCURY.INI generated by Mercury Setup # # #  Sample Bindery Mode MERCURY.INI file (in fact, the one I use on #  my production server). This one implements most of the possible #  switches for each module, but you can edit it to do less if you #  wish. All the fields shown below are described in the file #  MGUIDE.HLP which is supplied in the Mercury archive. # #  Anything after a '#' to the end of the line is a comment and #  is stripped out before parsing. Trailing and leading whitespace #  is also stripped before parsing. Many sample commands have been #  left in this file commented out for reference purposes. # [General] myname:          cpnt.gnb.ca    # Canonical name for this server timezone:        +0000    # Time Zone to add to date fields file_api:        1    # Use the file api instead of queues mailqueue:       D:\MERCURY\QUEUE    # Where mail should be put for delivery smtpqueue:       D:\MERCURY\QUEUE    # Where the SMTP client should look for mail newmail_path:    D:\MERCURY\MAIL\~N    # Where to find the users' WinPMail mailboxes. [Protocols] D:\MERCURY\MERCURYS.DLL D:\MERCURY\MERCURYP.DLL # D:\MERCURY\MERCURYE.DLL D:\MERCURY\MERCURYC.DLL # D:\MERCURY\MERCURYD.DLL # D:\MERCURY\MERCURYH.DLL # D:\MERCURY\MERCURYF.DLL # D:\MERCURY\MERCURYW.DLL # D:\MERCURY\MERCURYX.DLL # D:\MERCURY\MERCURYI.DLL [Mercury] failfile:      D:\MERCURY\Mercury\FAILURE.MER    # Delivery failure notification template confirmfile:   D:\MERCURY\Mercury\CONFIRM.MER    # Delivery confirmation template aliasfile:     D:\MERCURY\Mercury\ALIAS.MER    # System-wide alias file synfile:       D:\MERCURY\Mercury\SYNONYM.MER    # User synonym database listfile:      D:\MERCURY\Mercury\LISTS.MER    # List of lists logfile:       D:\MERCURY\Mercury\system.log    # Traffic logging file logwidth:      30    # Width of address fields in logfile retpath:       1    # Do/Don't use 'return-path' for errors maxhops:       30    # Used to detect mailing loops gullible:      0    # If NZ, don't validate 'From' fields bitnethost:    cunyvm.cuny.edu    # Relay host for '.bitnet' rewrites poll:          10   # Seconds between queue polling cycles scratch:       D:\MERCURY\Scratch    # Where we can write temp files returnlines:   15    # How many lines of failed messages to return postmaster:    depcon    # NetWare UIC of postmaster broadcast:     1    # Yes, we want broadcast notifications, but receipts:      0    # ... no broadcasts for receipt confirmations PM_notify:     1    # Do/Don't send errors to the postmaster change_owner:  1    # Change message ownership to recipient auto_tzone:    1    # If NZ, obtain timezone information from OS LogLevel:      25    # System message reporting level LogMax:        100    # Maximum number of system messages to store RetryPeriod:   5    # Period between job retries (minutes) MaxRetries:    30    # Maximum number of times a job can fail TwoPasses:     1    # If NZ, process submission queue in 2 passes Autoaddress:   0    # If NZ, auto-generate user addresses based on name Daily_exit:    0    # If NZ, auto-restart each day after maintenance No_Areply:     0    # If NZ, disable automatic replies system-wide Alt_Forward:   1    # If NZ, allow alternate forwarding specification Maint_hour:    2    # Hour for daily maintenance (24-hour clock) Maint_min:     0    # Minute for daily maintenance Retry_Mode:    0    # If NZ, use progressive backoff for queue retries Local_DSNs:    1    # If NZ generate Delivery Status for local users only DSN_time1:     3600    # Seconds deferred before first DSN is sent DSN_time2:     7200    # Seconds deferred before second DSN is sent DSN_time3:     14400    # Seconds deferred before third DSN is sent Host_in_title: 0    # If NZ, show hostname in system tray and taskbar Lingering:     0    # Enable or disable 'lingering mailboxes' Linger_Timeout: 60   # Seconds before a lingering mailbox is shut down Alert_Host:    notify.pmail.com    # Host to check for licensee advisories Alert_Interval: 720   # Frequency in minutes (min 240) for checks Alert_Flags:   3    # Flags controlling advisory checks [MercuryC] Host : mailserv.gnb.ca Failfile : D:\MERCURY\Mercury\FAILURE.MER HELO : CPNT.GNB.CA Poll : 30 Scratch : D:\MERCURY\Mercury ReturnLines : 15 Timeout : 30 ESMTP : 1 POP3_Auth : 0 Logfile : d:\mercury\logs\smtpcli.log Log_Verbose : 0 Session_logging : d:\mercury\logs\smtpcli Session_logmode : 0 [MercuryE] Poll : 15 Timeout : 60 # Session_logging : o:\mail\mercury\melogs # Session_logmode : 1 # Nameservers : 139.80.64.1 [MercuryD] scratch              :     D:\MERCURY\Mercury # Timeout : 30 # Poll : 120               # How often to check hosts for new mail (seconds) [MercuryS] Debug : 1 HELO : cpnt.gnb.ca Logfile : D:\MERCURY\Logs\MERCURYS.LOG Timeout : 30 Relay : 0 Strict_Relay : 0 Allow_Illegals : 0 SMTP_Authentication : 1 Auth_File : smtpauth Session_logging : D:\MERCURY\Logs\smtp.LOG Session_logmode : 0 Compliance_Settings : 0 Maximum_Failed_Rcpts : 4 Max_Relay_Attempts : 4 SSL_Mode : 0 ST_Blacklisting : 288 No_VRFY : 0 SMTP_ConnFlags : 0 [MercuryP] Scratch : D:\MERCURY\Mercury Logfile : pop3log.log Stack : 32768 Mark_Read : 1 SSL_Mode : 0 Login_Disabled : 0 UIDL_nul_list : 1 New_UIDs : 0 No_NUL_passwords : 1 Timeout : 60 [MercuryX] # Cmd_Delay: 10 # Cmd_Wait: 1 # IE4_Dialling: 0 # Use_ETRN: 1 # Clients_only: 1 # Sunday: 0800,1800,3,15,30,60 # Monday: 0830,2100,5,15,60,60 # Tuesday: 0830,2100,5,15,60,60 # Wednesday: 1500,1900,2,30,5,30 # Thursday: 0900,1900,2,30,0,0 # Friday: 0830,2100,5,15,60,60 # Saturday: 1031,1035,2,45,-1,0 # [Groups] # testgroup            :     TESTGRP [Domains] cpnt: cpnt cpnt: cpnt.gnb.ca [Maiser] Maiser:          Maiser   # 'Username' of mail server account Helpfile:        D:\MERCURY\Mercury\MAISER.HLP   # Help sent on failed maiser requests Lookupfile:      D:\MERCURY\Mercury\MAISER.LKP   # Format file for the 'lookup' command Send_dir:        D:\MERCURY\Mercury\SENDABLE   # Directory for the 'send' command Logfile:         D:\MERCURY\Logs\MAISER.LOG   # Mail server log file Notify:          D:\MERCURY\Mercury\TMP   # Dir. for pending 'notify' commands NoList:          N   # If 'Y', disable the 'list' command Local_only:      Y   # If 'Y', local 'notify' commands only [MercuryH] Logfile              :     D:\MERCURY\Logs\MERCURYH.LOG # Addressbook : \\CLIO\SYS\SYSTEM\MERCURY\PH.PMR # MOTD : \\CLIO\SYS\SYSTEM\MERCURY\PH-MOTD.TXT # Administrator : postmaster@clio.pmail.gen.nz Timeout : 30 [Groups] [Rewrite] [Statistics] StatFlags:    16  # Statistical reporting settings STF_Hours:    24  # How often to write stat log files STM_Hours:    24  # How often to e-mail stat log files </p>

<p>here is a ProcessExplorer screen while hung:</p><p> </p><p><img src="http://i42.tinypic.com/20sc6jo.jpg" title="Process Explorer screen print" alt="Process Explorer screen print" mce_src="http://i42.tinypic.com/20sc6jo.jpg" height="1600" width="1600"> </p>

<p>Do you have any daemons active (like ClamWall)? Any content control testing enabled? Is there any real-time antivirus program running on the server? Any other programs or services that might access the same folders as Mercury?</p><p>/Rolf </p>

<p>no daemons, no filtering, but we are running the Windows firewall with holes for ports 25 and 110. Also we were running McAfee AV and now I have turned the File Access Scan off as it I did suspect it could be an issue.</p><p> </p><p>On a related note, my 3rd party SMTP client, written in DOT-NET has encountered the issue descirbed in this link:</p><p> </p><p><a href="http://connect.microsoft.com/VisualStudio/feedback/details/146711/smtpclient-does-not-gracefully-close-the-underlying-tcp-ip-connection#details" title="http://connect.microsoft.com/VisualStudio/feedback/details/146711/smtpclient-does-not-gracefully-close-the-underlying-tcp-ip-connection#details" mce_href="http://connect.microsoft.com/VisualStudio/feedback/details/146711/smtpclient-does-not-gracefully-close-the-underlying-tcp-ip-connection#details">http://connect.microsoft.com/VisualStudio/feedback/details/146711/smtpclient-does-not-gracefully-close-the-underlying-tcp-ip-connection#details</a></p><p> </p><p>Is it possible this is causing the issue?</p><p> </p><p>Peter </p>

<p>With the right ports open Windows firewall won't cause any problems. McAfee AV on the other hand could definitely interfere. </p><p>If the misbehaving .net component fails to close connections properly you should see some trace of this in the logs. It wouldn't cause a lot of processor load for Mercury, though, unless perhaps if there was a big number of simultaneous connections from that client. At some point Windows would run out of available TCP connections as well, presumably.</p><p>/Rolf </p>

<p>no recurrence of the issue since turning off McAfee File Access Scan.</p><p> </p><p>Peter[:)] </p>