Re: tbtape -c in background

At 01:46 PM 4/15/97 +0000, Mark D. Stock wrote: >Dwight Rudolph Jr wrote: >> >> In article <334DEF28.20DC@west.co.za>, marks@west.co.za says... >> > >> >Greg Salmon wrote: >> >> >> >> Mark D. Stock wrote: >> >> > >> >> > Neil Truby wrote: >> >> > > >> >> > > I've been asked to set up a background script to automatically run a >> >> > > continuous logical log backup when my UNIX server starts. >> >> > >> >> > This can be done, but is SO dangerous that I'm not even going to >> >> > say how. I would only implement this idea if I could ensure an >> >> > operator was present 24 hours a day to pull out the tape when the >> >> > system crashes. Then of course they might as well do a manual one >> >> > if they are sitting there anyway. :-) > >> Actually the whole log backup process sort of irritates me. We do a full >> level 0 backup every night and we're willing to reload that backup (losing a >> day's work) if there's a disk crash. So the logical logs aren't worth all >> that much to us. >> >> I realize there are lots of places that must recover every committed >> transaction, but I'm sure there are lots, like us, that don't. An option to >> NOT do logical logs (like Rdb/VMS has) would be really welcome. > >There are two: > >1) Switch off logging at the database level. This assumes you > are running an inteligent application that can detect and deal > with this change. > >2) Set your logical log backup device to /dev/null. This is picked > up as a special device and logs are freed automatically. > >Help that helps, >-- > Mark. This is a script I use to start the ontape -c process in the background whenever the instance's are restarted or after I have perform an Archive of the online instance. I keep a week of logical logs on disk. Which are dated and have a unique name I also backup these logs to tape before an Archive each night. The only problem one might have is If .. your logical logs grow in the course of a day greater than the space you have in the file system on which they are stored. I have been using this for about 6 month in a production and development env. and have had to restore several times from a level 1 archive with no problem # "@(#)startlog.sh - run informix ontape -c" EXITOK=0;SIGHUP=1;SIGINT=2;SIGTERM=1;SIGQUT=3 TrapSigs="$EXITOK $EIGHUP $SIGINT $SIGTERM $SIGQUT" trap 'rm -f tmp$$ > /dev/null 2>&1; exit 0 ' $TrapSigs mail_list="informix jallen" PWD="/hp/informix/admin/logs" PROGDIR="/hp/informix/admin" engine_file="$PROGDIR/scripts/online.engines" engines=`grep -v "^#" $engine_file | cut -d ":" -f1` DATE=`date` dayofweek=`date +%u` BUDIR=/hp/informix/admin/backup error_file=${BUDIR}/arch_error ARCHIST=${BUDIR}/arch_history # stop any ontape -c process which is running pid=`ps -ef | grep "ontape -c" | grep -v "grep" | cut -c9-14` if [ "$pid" != "" ] then ps -ef | grep "ontape -c" | grep -v "grep" | cut -c9-14 | kill -15 $pid fi arch_logs() { dayofweek=`date +%u` olddate=`expr $dayofweek - 1` if [ "$olddate" = 0 ] then olddate=7 fi # compress yesterdays logs OLDLOG=${BUDIR}/logs/L$instance$olddate compress $OLDLOG # null out logs that are one week old dayofweek=`date +%u` LEVLOG="/hp/informix/admin/backup/logs/L$instance$dayofweek" if test -w $LEVLOG.Z then cp -f /dev/null $LEVLOG chmod 666 $LEVLOG rm -f $LEVLOG.Z fi LEVLOG=${BUDIR}/logs/L$instance$dayofweek #Set LTAPEDEV in onconfig to a unique name the this day. cat $INFORMIXDIR/etc/$ONCONFIG | awk ' { if ($1 == "LTAPEDEV") $2=backarc printf("%s\\n",$0) }' backarc=$LEVLOG > $INFORMIXDIR/etc/tblogtmp.int cp -f $INFORMIXDIR/etc/$ONCONFIG $INFORMIXDIR/etc/$ONCONFIG"bk" mv -f $INFORMIXDIR/etc/tblogtmp.int $INFORMIXDIR/etc/$ONCONFIG echo "Restarting log archiver..." >> $ARCHIST echo $LEVLOG >> $ARCHIST nohup $INFORMIXDIR/bin/ontape -c < $PWD/ontape_cr 1> /dev/null 2>&1 & echo "restarted" } # get the list of instances on my platform list=$engines # for each instance start ontape -c for instance in $list do ONCONFIG=`grep -v "^#" $engine_file | grep $instance | cut -d ":" -f2` DBSERVERNAME="$instance" INFORMIXSERVER="$instance" export ONCONFIG export DBSERVERNAME export INFORMIXSERVER echo "Begin Logical log backup at `date` for $instance" >> $ARCHIST case $instance in devl_hp1) arch_logs ;; onlprod) arch_logs ;; esac # if any errors mail informix and dba if [ -s $error_file ]; then mail $mail_list < $error_file cp /dev/null $error_file fi O O O O O O . . . ______________________________ __======______T___ O ____ ||Joseph Allen | |H (602)582-1064 | .] [__n_n_|DD[ ===___ |Email jsa14@worldnet.att.net| |W (602)631-7592 | >(_________|__|__[_____]_|Fax (602) 582-1064 _|_ |Sr.DBA CSK Auto_|_ _/oo OOOOO oo' oo oo 'o^o^o o^o^o' 'o^o o^o' +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+--+-+