Re: PMAS PMDF channel log files overflowing file version number - VMS

This is a discussion on Re: PMAS PMDF channel log files overflowing file version number - VMS ; I recently added queue keywords to the CONVERSION and PMAS channels. so that they'd run in their own queues and not block the delivery jobs. I just discovered that ALL the mail batch queue entries, including those for the PMAS ...

+ Reply to Thread
Results 1 to 3 of 3

Thread: Re: PMAS PMDF channel log files overflowing file version number

  1. Re: PMAS PMDF channel log files overflowing file version number

    I recently added queue keywords to the CONVERSION and PMAS channels.
    so that they'd run in their own queues and not block the delivery jobs.

    I just discovered that ALL the mail batch queue entries, including those
    for the PMAS and CONVERSION jobs, include /LOG qualifiers.

    What might be causing that?

    David Webb wondered

    > Have you got master_debug defined on those channels (or on the defaults line) ?


    master_debug used to be set, but it was removed some time ago
    in an attempt to improve througput.
    The defaults currently set in PMDF.CNF are

    defaults noswitchchannel logging notices 1 3 6 9 12

    Would the "logging" keyword be the culprit now?
    I do need to have the channel status messages logged in MAIL.LOG_CURRENT

    The individual channels in PMDF.CNF have no logging or debug entries.
    e.g.

    conversion defragment queue mail$conv
    CONVERSION-DAEMON

    pmas queue mail$pmas
    PMAS-DAEMON

    Aha!

    The batch queue entries all include /LOG qualifiers.
    What would be causing that?


    All of the debug options listed in OPTION.DAT are set to 0
    Of its LOG_* options, the only ones set to non-zero values are

    LOG_SNDOPR=1
    LOG_MESSAGE_ID=1
    LOG_FORMAT=2
    LOG_CONNECTION=3
    LOG_DEQ=1000
    LOG_STATISTICS=1
    LOG_SIZE_BINS=2,10,50,100,500
    LOG_DELAY_BINS=60,600,6000,60000,600000

    > If not then I'd have a look at the contents of those log files since I'd
    > suspect it indicates some error is occurring.


    The *MASTER*.LOG files all contain only the default system login and
    logout messages.

    > I am not seeing gigantic numbers of


    > CONVERSION_MASTER.LOG
    > or
    > PMAS_MASTER.LOG



    > files.



    > >Usually it takes several attempts to get rid of them all so the
    > >log file version number can be reset to 1, since the rate of incoming
    > >messages usually keeps several recent, high-version-number log files
    > >locked for a while -- which presumably is why PMDF can't eliminate them
    > >on its own.
    > >

    > You should be able to rename the locked files.


    Thanks for reminding me.
    I've updated my cleanup script to rename the log files to
    a subdirectory and then delete them a day later.
    I guess that'll do for now.

  2. Re: PMAS PMDF channel log files overflowing file version number

    In article <01MN69TZ7S4E8Y4WU3@LNS62.LNS.CORNELL.EDU>, Selden E Ball Jr writes:
    >I recently added queue keywords to the CONVERSION and PMAS channels.
    >so that they'd run in their own queues and not block the delivery jobs.
    >
    >I just discovered that ALL the mail batch queue entries, including those
    >for the PMAS and CONVERSION jobs, include /LOG qualifiers.
    >
    >What might be causing that?
    >
    >David Webb wondered
    >
    >> Have you got master_debug defined on those channels (or on the defaults line) ?

    >
    >master_debug used to be set, but it was removed some time ago
    >in an attempt to improve througput.
    >The defaults currently set in PMDF.CNF are
    >
    >defaults noswitchchannel logging notices 1 3 6 9 12
    >
    >Would the "logging" keyword be the culprit now?


    No my defaults line is

    defaults noswitchchannel logging nosendpost nowarnpost inner immediate immnonurgent notices 1 2 4 6


    >I do need to have the channel status messages logged in MAIL.LOG_CURRENT
    >
    >The individual channels in PMDF.CNF have no logging or debug entries.
    >e.g.
    >
    >conversion defragment queue mail$conv
    >CONVERSION-DAEMON
    >
    >pmas queue mail$pmas
    >PMAS-DAEMON
    >
    >Aha!
    >
    >The batch queue entries all include /LOG qualifiers.
    >What would be causing that?
    >

    On my systems the batch queues are created as

    $ initialize/queue/device=server/noenable_generic/processor=pmdf_process_smb-
    /protection=(s:rwe,o:rwd, g:r, w:r) mail$batch_imhub1_1

    etc

    with the queues being fed by generic queues such as

    $initialize/queue/device=server/generic=(mail$batch_imhub1_5, -
    mail$batch_imhub1_6, mail$batch_imhub2_5, -
    mail$batch_imhub2_7, mail$batch_imhub1_8,mail$batch_imhub2_9, -
    mail$batch_imhub2_6,mail$batch_imhub1_7, mail$batch_imhub1_17, -
    mail$batch_imhub2_17, mail$batch_imhub1_18, mail$batch_imhub2_18, -
    mail$batch_imhub1_19, mail$batch_imhub2_19, mail$batch_imhub1_20, -
    mail$batch_imhub2_20, -
    mail$batch_imhub1_26, mail$batch_imhub2_26, mail$batch_imhub1_27, -
    mail$batch_imhub2_27, mail$batch_imhub1_28, mail$batch_imhub2_28, -
    mail$batch_imhub2_31, mail$batch_imhub1_31, -
    mail$batch_imhub2_35, mail$batch_imhub1_35, -
    mail$batch_imhub2_36, mail$batch_imhub1_36, -
    mail$batch_imhub2_37, mail$batch_imhub1_37, -
    mail$batch_imhub2_38, mail$batch_imhub1_38, -
    mail$batch_imhub2_39, mail$batch_imhub1_39, -
    mail$batch_imhub2_40, mail$batch_imhub1_40, -
    mail$batch_imhub2_41, mail$batch_imhub1_41 -
    ) /protection=(s:rwe,o:rwd, g:r, w:r) mail$convert


    And show/que/all/full shows entries like


    Server queue MAIL$BATCH_IMHUB1_15, busy, on IMHUB1::, mounted form DEFAULT
    /BASE_PRIORITY=4 /DEFAULT=(FEED,FORM=DEFAULT) /NOENABLE_GENERIC
    /OWNER=[SYSTEM] /PROCESSOR=PMDF_PROCESS_SMB /PROTECTION=(S:RSM,O:RSD,G:R,W:R)

    Entry Jobname Username Blocks Status
    ----- ------- -------- ------ ------
    475 PMDF channel PMAS master
    SYSTEM 82 Processing
    Submitted 31-OCT-2007 18:14:21.47 /FORM=DEFAULT /PARAM=("PMAS",
    "NOPOLL","31-OCT-2007:18:14:18.07","01MN6KKK5HEA8Y65BR","","","2:999")
    /PRIORITY=100
    File: _$255$DKD1:[PMDF.COM]MASTER.COM;3 (processing))


    I do have a few

    TASK_SERVER_MAIL$BATCH_IMHUB1_1.LOG etc files which just contain

    $ Set NoOn
    $ VERIFY = F$VERIFY(F$TRNLNM("SYLOGIN_VERIFY"))

    but not a gigantic number.

    The only other log files being produced in any number are

    TCP_LOCAL_SLAVE.LOG files containing

    29-OCT-2007 09:58:18.01: SMTP routine failure from SMTPC_ENQUEUE

    and a few

    TCP_SMTP_SERVER.LOG files containing

    $ Set NoOn
    $ VERIFY = F$VERIFY(F$TRNLNM("SYLOGIN_VERIFY"))

    >
    >All of the debug options listed in OPTION.DAT are set to 0
    >Of its LOG_* options, the only ones set to non-zero values are
    >
    >LOG_SNDOPR=1
    >LOG_MESSAGE_ID=1
    >LOG_FORMAT=2
    >LOG_CONNECTION=3
    >LOG_DEQ=1000
    >LOG_STATISTICS=1
    >LOG_SIZE_BINS=2,10,50,100,500
    >LOG_DELAY_BINS=60,600,6000,60000,600000
    >


    My settings in option.dat for log entries are

    Imhub2:search pmdf_tableption.dat log
    SEPARATE_CONNECTION_LOG=0
    LOG_SNDOPR=1
    LOG_MESSAGE_ID=1
    LOG_SENSITIVITY=0
    LOG_FILENAME=1
    LOG_NOTARY=0
    LOG_USERNAME=1
    LOG_FORMAT=1
    LOG_LOCAL=0
    LOG_CONNECTION=3
    LOG_ALQ=250000
    LOG_DEQ=20000
    LOG_DEBUG=0
    LOG_NODE=0
    LOG_PROCESS=0
    LOG_HEADER=0
    LOG_STATISTICS=1
    LOG_CONNECTIONS_SYSLOG=0
    LOG_MESSAGES_SYSLOG=0
    LOG_SIZE_BINS=2,10,50,100,500
    LOG_DELAY_BINS=60,600,6000,60000,600000


    David Webb
    Security team leader
    CCSS
    Middlesex University


    >> If not then I'd have a look at the contents of those log files since I'd
    >> suspect it indicates some error is occurring.

    >
    >The *MASTER*.LOG files all contain only the default system login and
    >logout messages.
    >
    >> I am not seeing gigantic numbers of

    >
    >> CONVERSION_MASTER.LOG
    >> or
    >> PMAS_MASTER.LOG

    >
    >
    >> files.

    >
    >
    >> >Usually it takes several attempts to get rid of them all so the
    >> >log file version number can be reset to 1, since the rate of incoming
    >> >messages usually keeps several recent, high-version-number log files
    >> >locked for a while -- which presumably is why PMDF can't eliminate them
    >> >on its own.
    >> >

    >> You should be able to rename the locked files.

    >
    >Thanks for reminding me.
    >I've updated my cleanup script to rename the log files to
    >a subdirectory and then delete them a day later.
    >I guess that'll do for now.


  3. Re: PMAS PMDF channel log files overflowing file version number


    > Would the "logging" keyword be the culprit now?


    The logging keyword has nothing to do with the creation of batch log
    files. It controls whether information on message enqueue/dequeue
    operations are written to the MAIL.LOG_CURRENT file.

    > The individual channels in PMDF.CNF have no logging or debug entries.
    > e.g.
    >
    > conversion defragment queue mail$conv
    > CONVERSION-DAEMON


    But if logging were on the defaults line, it would be implicitly applied
    to all channels below it.

    ---------------------------------------------------------
    Tom Wade | EMail: tee dot wade at eurokom dot ie
    EuroKom | Tel: +353 (1) 296-9696
    A2, Nutgrove Office Park | Fax: +353 (1) 296-9697
    Rathfarnham | Disclaimer: This is not a disclaimer
    Dublin 14 | Tip: "Friends don't let friends do Unix !"
    Ireland

+ Reply to Thread