[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference rocks::dec_edi

Title:DEC/EDI
Notice:DEC/EDI V2.1 - see note 2002
Moderator:METSYS::BABER
Created:Wed Jun 06 1990
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:3150
Total number of notes:13466

3055.0. "Fetch delay" by UTRTSC::SMEETS (Workgroup support) Thu Mar 13 1997 14:21

T.RTitleUserPersonal
Name
DateLines
3055.1Try this work-around...SCASS1::EDITEX::GONYEAThu Mar 13 1997 20:5121
    Martin,
    
    This is a known problem.  We've experienced it at a number of sites. 
    I don't believe that the problem has been fixed, but there is a
    work-around that we've used that seems to resolve the problem.  There
    is a logical DECEDI$AFS_QUE_REFRESH which defines how often the search
    list is updated in CACHE (my technical explanation may be incorrect,
    but this is the "laymens" description!). If you reduce this (I believe
    the default is 6 hours) cache will be updated more often and the search
    list will hopefully contain all documents.  Mark Thompson can certainly
    provide you with a better technical explanation and let you know if
    this has been fixed or what the outlook is.  One note....my
    understanding is that lowering this can have a negative impact on
    performance, particularly in high-volume shops.
    
    Hope this helps.
    
    Paul Gonyea
    
    Electronic Commerce Practice Group
    Dallas TX
3055.2FORTY2::DALLASPaul Dallas, DEC/EDI @REO2-F/E2Fri Mar 14 1997 12:1111
    This won't be fixed until V3.2.
    
    The problem is caused by a change in the caching algorithm to cope with
    high-volume sites. The AFS used to clear its cache and reload every
    time it failed to find a requested document. Now it merely does an
    incremental update and only flushes the cache when the refresh timer
    expires. The problem is that some documents can miss this incremental
    update, because they take a longer time to translate. 
    
    The problem is very common if you run multiple translators, since the
    documents can easily get to AVAILABLE out of sequence.