Continuos logging procedures

["Shulzhenko Vasyl" <vasilis@softline.kiev.ua>]

>
>Rogers Patterson wrote in message <7c707i$t3k$1@news.xmission.com>...
>>
>>We recently started doing continuos logging to tape. However, we have not
>>come up with a good procedure to automate this process. We currently have a
>>DBA start the logging at the beginning of the workday and then end of the
>>day he/she interrupts the process, goes across town to the server, and
>>labels the tape.
>
>
>How many logical logs do you fill up in a day ?? If you don't know, modify
>the log_full.sh program to echo a line to a file every time a log fills ( an
>example is given below...)
>--------------------------------------------
>EVENT_SEVERITY=$1
>EVENT_CLASS=$2
>EVENT_MSG="$3"
>EVENT_ADD_TEXT="$4"
>EVENT_FILE="$5"
>
>ALARM_FILE=$SA_HOME/ifal_alarm.out   ё changed 26/10/98
>DAILYLOGS=/home/informix/etc/dailylogs
>
>case "$EVENT_CLASS" in
>        23)
>                ё Record each log that gets filled in a file which
>                ё is created on a daily basis, so that we can monitor
>                ё the number of logical logs that get filled each day.
>
>                DATE=`date +%d%m%y`
>                echo $EVENT_MSG >> $DAILYLOGS.$DATE
>
>                ё onbar assumes no operator is present,
>                ё so all messages are written to the activity
>                ё log and there shouldn't be any output, but
>                ё send everything to /dev/null just in case
>
>                if [ -f /tmp/stoplogs ]
>                then
>                  echo $EVENT_SEVERITY >>$ALARM_FILE
>                  echo $EVENT_CLASS >>$ALARM_FILE
>                  echo "Archive Tapes being changed. Archiving of logs has
>been
>temporarily suspended" >>$ALARM_FILE
>                  echo " " >>$ALARM_FILE  ё Create 6 lines per incident
>                  echo " " >>$ALARM_FILE
>                  echo `date` >>$ALARM_FILE
>                  EXIT_STATUS=0
>                else
>                  $BACKUP_CMD 2>&1 >> /dev/null
>                  EXIT_STATUS=$?
>                fi
>                ;;
>
>ё One program is shared by all event alarms.  If this ever gets expanded to
>ё handle more than just archive events, uncomment the following:
>        *)
>                case "$EVENT_SEVERITY" in
>           5)
>                echo $EVENT_SEVERITY >>$ALARM_FILE
>                echo $EVENT_CLASS >>$ALARM_FILE
>                echo $EVENT_MSG >>$ALARM_FILE
>                echo $EVENT_ADD_TEXT >>$ALARM_FILE
>                echo $EVENT_FILE >>$ALARM_FILE
>                echo `date` >>$ALARM_FILE
>                EXIT_STATUS=1
>            ;;
>        4)
>                echo $EVENT_SEVERITY >>$ALARM_FILE
>                echo $EVENT_CLASS >>$ALARM_FILE
>                echo $EVENT_MSG >>$ALARM_FILE
>                echo $EVENT_ADD_TEXT >>$ALARM_FILE
>                echo $EVENT_FILE >>$ALARM_FILE
>                echo `date` >>$ALARM_FILE
>                EXIT_STATUS=1
>            ;;
>        3)
>                echo $EVENT_SEVERITY >>$ALARM_FILE
>                echo $EVENT_CLASS >>$ALARM_FILE
>                echo $EVENT_MSG >>$ALARM_FILE
>                echo $EVENT_ADD_TEXT >>$ALARM_FILE
>                echo $EVENT_FILE >>$ALARM_FILE
>                echo `date` >>$ALARM_FILE
>                EXIT_STATUS=1
>            ;;
>        *)
>            EXIT_STATUS=0
>            ;;
>        esac
>
>                EXIT_STATUS=1
>        ;;
>esac
>
>exit $EXIT_STATUS
>---------------------------------------------
>
>If you aren't filling up that many ( 100 - 200 ) why not just create 300
>logs then do a daily archive of the logical logs - we also have a script
>that monitors the number of logs that fill up and when it gets to 75%, an
>e-mail is raised to all support and admin staff so that the necessary action
>can be taken.
>
>>
>>Does anyone out there have a procedure worked out that keeps continuos
>>logging running the entire time the instance is up, is automatically
>>restarted after a crash of the instance or Unix box, and is able to provide
>>the log ranges on the tape to an operator that can label and switch tapes?
>
>
>We use onbar with the log_full.sh program ( as described above ). It works
>fine and have not had any (major) problems - granted, we are using Omniback
>and not ISM as we are using 7.24.UC7. You should definitely consider using
>onbar - ontape is fine but, as you said, it can be a pain when you have to
>stop and re-start it.
>
>A few good steps would be:-
>1. Write a script to monitor the number of logs that get filled on a daily
>basis
>2. Write a script that when the logs are 75% full, creates a warning of some
>description
>3. Start using the log_full.sh program for what it was designed for
>4. configure onbar to archive the logs to a specific tape device, used
>solely for archiving the logical logs - so a tape drive costs $4000 - ask
>your FD how much his data is worth.
>
>>
>
>>As you have probably suspected, we do not have any tape storage manager
>>software at this time.
>
>
>Not strictly true with the 7.3 release, you have ISM (Informix Storage
>Manager) which can be configured to perform onbar archives of the logical
>logs through the log_full.sh alarm program.
>
>>
>>TIA
>>Rogers Patterson
>>
>>Ps. We are IDS 7.3 on DEC Unix 4.0D
>
>
>Hope this helps
>
>Sean
>
>
>