Conference iosg::all-in-1_v30

    More information:
    
    VMS Version 5.4-2, ALL-IN-1 2.4, patched to K604.
    
    Entry in MTILOG.LOG says that only one local delivery was made where in
    fact 3 other people received it.
    
    Checked to make sure that there were no problems with nicknames,
    distribution lists, mail forwarding etc.  Doing a SH on the message
    points to one addressee only and she did not receive the mail).
    
    Entries in MTIERR.LOG for the time the message was sent are as follows:
    
    27-JUL-1992 16:09:23           %OA-E-RECORD_IO_ERROR, Error occurred
    while attempting to UPDATE the record
    27-JUL-1992 16:09:24           -RMS-E-EXT, ACP file extend failed
    27-JUL-1992 16:23:12           %OA-E-RECORD_IO_ERROR, Error occurred
    while attempting to SHARE the record
    27-JUL-1992 16:23:12           -RMS-E-EXT, ACP file extend failed
    
    Any ideas please.....
    
    Sandra
    [EOB]
    

    there was a complicated bug whereby if it the sender crashed, it could 
    leave in memory the previous addresses - it was fixed, but don't know
    which patch it went into...
                               
    Alternaltively it could be a completely different problem.
                               
    Winton

    Winton,
    
    I don't see a fix for a Sender problem such as the one you describe
    in any K6xx patch perhaps you could give us a little more of a clue.
    
    Anyway it probably doesn't matter since .0 has K604 installed and
    since K605 only included WPS-PLUS fixes the system is fully-patched
    as far as the Sender/Fetcher are concerned.
    
    So it must be another problem.
    
    Regards,
    
    Andrew.D.Wicks  

I have had a look at this problem and it does appear that ALL-IN-1 has mailed
people incorrectly. The user was in fact forwarding a received message to one
user and it arrived at three others. The recipiants do not appear on the original
message and I can find no link between them (names OVERTON_M, MATTHEWS_P, 
REINDORPP_P when should have been delivered to BUCKLEY_M.

Has anybody any further idea as to what could cause this as we are concerned of
the security implications of further occurances.

Relevant records from site:

>OA$MTI_ERR
>27-JUL-1992 16:09:23           %OA-E-RECORD_IO_ERROR, Error occurred while 
>attempting to UPDATE the record
>27-JUL-1992 16:09:24           -RMS-E-EXT, ACP file extend failed              

The disks have not been defragmented for several weeks. I assume that this could
be SDAF, PENDING or INDEXF failing to extend at the time of this send. The 
unusual thing is that the sender of the message was one of only a handful of 
users at this time writing into this particular SDAF (the other 600 users 
writing  to DAF_E on a different disk).

>OA$MTI_LOG
>27-JUL-1992 16:10:14 EVANS_A    N    N    1     0      OA$SHARD648:ZUGDN4EM8.WPL

This says only one local addressee when actualy three users were delivered it.
Do these values come from the header of the message or the actual number of 
people delivered to?

CABFIX dump of SDAF record. The delivered date has a duplicate value but this
shouldnt cause a problem. Notice Usage count is 1 when three people are 
referencing it at present (nb I believe FCVR hasn't been run for a good while)

DAF_KEY='OA$SHARD648:ZUGDN4EM8.WPL                                      '
USAGE_COUNT=    1  UNUSED='  ..    '  TOTAL_BYTE_SIZE=  679

RECORD_CODE='  '  CONT_FLAG='.'  THIS_REC_ATTR_SIZE=  675
DR='NO'
RR='NO'
PRIORITY='FIRST_CLASS'
FORWARDABLE='YES'
TO='MYRA BUCKLEY                         ( BUCKLEY_M )'
SUBJECT='This is the arlier attchment I referred to'
FROM_ADDRESS='EVANS_A'
FROM_DEPT='S.I. Business Management        '
DAUTHOR='Arthur Evans                  '
FROM_TEL_NO='(7)782 2132                     '
DELIVERED='1992072716100000'
DELIVERED='1992072716100000'
POSTED='1992072716100000'
ATT_DOC='OA$SHARB833:ZUFYJT6Y4.WPL'
DTITLE='This is the arlier attchment I referred to
'
DTYPE='MAIL           '
DSETUP='NONE           '
DCREATED='1992072716090000'
DFORMAT='               '
DMODIFIED='1992072716090000'
DDSAB='WPSPLUS        '
LANGUAGE='BRITISH'

I'd suggest raising an SPR for this (no, it doesn't help your customer, but 
tidying up the discs to give more room to SDAFs and Pending might). There is 
definitely a problem in the error handling code here, and it can only be 
properly investigated if someone goes through the whole path carefully to 
ensure all holes are plugged - and that probably won't happen unless there's 
an offcial bug to fix.

grahame

    re: which patch kit it was in, dunno, after I fix'em, i forget them! 
                                                                      
    Seriously, I never know the fate, after they leave our TLC. 

    I Do Remember it was in response to a CLD from an Irish Bank.
                                            
    The original cause was never discovered, but the bug was caused if a
    message bombed out, and then the sender carried, on, because
    non-Volatile (OWN) memory  was being used. The sympton was that the
    next message along, would pickup the addressess from the last one.
                                                        
    Winton                                  

    There is some similarity with the problem and the one caused by the Sender
    crashing, as the people that received the mail here all work for the
    same group!  I have checked all accounts to see if I could find some
    commonality with previous mails sent by the person who sent the mail
    and all who received it, but there was nothing obvious...
    
    One thing does bother me though, how many other people may have
    received this message that we don't know about.  In this case the mail
    was not confidential, but sender of the original mail is a very senior
    manager who does have a lot of confidential mail in his account.
    
    Sandra 

    
     Similar problem I have (note #906) in TM. We cannot reproduce the
    problem. So far the customer said they're not sure if this was the
    first occurence of the problem.
    

    Mail being delivered to the wrong user on a customer site has happened
    at least 2 known times and maybe more in the past day.  The symptoms
    are as follows:
    
    1st instance:
    
    USERA sends mail to a remote USERB via X.400 gateway.  Mail gets
    delivered to USERB and is also delivered to an unrelated local USERX
    who is not on the address list.
    
    
    2nd instance:
    
    USERC sends mail to a remote USERD via X.400 gateway and to local
    USERE.  Mail gets delivered to recipients and also gets delivered to
    unrelated local USERX (same user as in instance 1) who is not on the
    address list.
    
    In both cases errors were logged in OA$MTI_ERR at around the same time:
    
    %OA-E-RECORD_IO_ERROR, Error occurred while attempting to UPDATE the
    record
    -RMS-E-RND, record not found
    
    $ ANAL/RMS on all shared DAF files shows no RMS errors.
    
    ALL-IN-1 V2.4, unpatched.
    
    Should we suspect PENDING.DAT corruption or SDAF.DAT corruption?
    
    I've never used CABFIX?  Would it be useful in this case?
    
    Any help appreciated,
    
    /Mario

    Mario,
    
    Apparently there's an open CLD on this - i don't want to post anymore
    here but contact me offline - if you get the CLD report it's one of
    the ones from the Basingstoke CSC.  
    
    if only i could keep up with my unread mail....
    
    Regards,
    
    Andrew.D.Wicks
    
    P.S for Sunil it was 00::06 here