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

Conference ulysse::rdb_vms_competition

Title:DEC Rdb against the World
Moderator:HERON::GODFRIND
Created:Fri Jun 12 1987
Last Modified:Thu Feb 23 1995
Last Successful Update:Fri Jun 06 1997
Number of topics:1348
Total number of notes:5438

676.0. "INGRES fast commit and database recovery." by PANIC::LILBURN (Angus 'Jocky' Lilburn) Fri Jun 22 1990 18:21

    Hello,
    	I have place this question on the ULTRIX/SQL notes conference as I
    am really interested in that fine Digital product, but the response has
    been nil, so since I believe the answer cannot be different, can any
    knowledgable INGRES specialist help me?
    	The subject is FAST COMMIT, which uses a cache to hold database
    updates before flushing them to disk.
    	My question is, if a system failure occurs when the cache if
    partially full,
    	i)	is recovery afterwards fully automatic
    or
    	ii)	does it require operator intervention to roll forward the
    		AIJ/Tx log since it was last flushed
    or
    	iii)	are the transactions lost?
    
    	Thanks for any constructive reponse,
    		Angus
T.RTitleUserPersonal
Name		DateLines
676.1watch this spaceWARNUT::BRYANFri Jun 22 1990 18:594
    Angus, the area of Ultrix/Sql's I/O management is being investigated by
    one of our UNIX guys. As soon as he gets anything sensible out of
    Ingres, which seems to be proving a little difficult, I will post it
    here.
676.2MerciPANIC::LILBURNAngus 'Jocky' LilburnMon Jun 25 1990 12:011
676.3try thisWARNUT::BRYANTue Jun 26 1990 13:483
    See entry 990 in the conference NAC::ULTRIX, I'm not sure that it
    provides all the answers your looking for but I think you will find it
    interesting reading.
676.4Saw it, read it, found itPANIC::LILBURNAngus 'Jocky' LilburnTue Jun 26 1990 14:594
    NAC::ULTRIX note 1454 is the note I think you mean, which in turn
    points to another. Found it anyway, thanks,
    	Angus
676.5my mistakeWARNUT::BRYANTue Jun 26 1990 15:206
    Uhm, OK, the author of note 1454 has supplied me with a hardcopy of
    edited highlights of the discussions around this area in the Ultrix
    conference. His highlights begin at 990, but most of the discussion is
    contained within note 1373, in fact he does not include anything
    from 1454. you might like to have a look at 1373.
676.6Le answerPANIC::LILBURNI'd rather be sleepingThu Aug 09 1990 12:315
    Hi Y'all,
    	Well the answer appears to be iii), you may lose transactions.
    Thanks to Joshua Friedman of Ingres Support for supplying the answer in
    note 124.3 of the ULTRIX_SQL conference,
    		Angus