Summary: Notification of DiskSuite 4.1 error message

From: Mark_Conroy@em.fcnbd.com
Date: Mon Aug 03 1998 - 15:07:32 CDT


     To all.
     
     Thanks for those who responded:
     
     John Speno
     Orn Asgeirsson
     MArk Baldwin
     Don Bodle
     
     
     I have included the scripts that I have received and again, many thanks.
     
     
     The following script is from Don Boddle:
     
     #!/usr/bin/sh
     # ------------------------------------------------------------
     #
     # File : meta-check
     #
     # Description : reviews output of metastat command for
     # hot spares in use.
     #
     # Author: Don Bodle, C.E.T.A.
     #
     # Date: 19 Dec 1997
     #
     # Notes: Placed in hcdeb's cron for checking periodically
     # --------------------------------------------------------------
     #
     /usr/opt/SUNWmd/sbin/metastat | /usr/bin/awk '
     BEGIN { found = 0 }
     /Hot Spare/ { found = NR + 1
     }
     { if ( NR >= found && found != 0) if ( $2 == "In" ) {
     printf("Hot Spare %s is %s %s \n",$1,$2,$3)
     }
     }' > /tmp/meta-check.tmp
     # originally just piped output to mailx, but would generate mail always.
     # to prevent mailing empty messages with just subject line, temp file was
     # used. If file isn't empty, send message. Otherwise just remove file.
     #
     if [ -s /tmp/meta-check.tmp ]
     then
       /usr/bin/mailx -c mes@mead.com -s"Metastat report" hcdeb@mead.com <
     /tmp/meta-
     check.tmp
     fi
     rm /tmp/meta-check.tmp
     
     
     
     
     
     
     The following script is from Mark Baldwin:
     
     #!/bin/sh
     
     if [ -f /var/log/md.log ]; then
     mv /var/log/md.log /var/log/md.log.old fi
     
     if [ -f /usr/opt/SUNWmd/sbin/metastat ]; then /usr/opt/SUNWmd/sbin/metastat
     |grep -i maint > /var/log/md.log elif [ -f /usr/etc/metastat ]; then
     /usr/etc/metastat |grep -i maint > /var/log/md.log fi
     
     if
     test -s /var/log/md.log; then
     mailx -s "`uname -n` metastat problem report" root < /var/log/md.log; else
     exit 0;
     fi
     
     
     
     
     
     The following script is from Orn Asgeirsson:
     
     This his already been covered, see the archives.
     
     I use this:
     
     0,30 * * * * /etc/ods_check > /dev/null
     
     
     #!/bin/sh
     # /etc/ods_check
     # mdcheck - Monitor the metatool devices and check its state
     
     PATH=/bin
     if [ ! -d /usr/opt/SUNWmd/sbin ]
     then
     echo ODS not installed.
     exit 1
     else
     PATH=$PATH:/usr/opt/SUNWmd/sbin
     fi
     WHOCARES=root@domain
     THISHOST="`uname -n`"
     #HALFHOUR=1800
     #while [ 0 -eq 0 ]
     #do
     FIELDS="`metastat|awk '{if ($1 == "State:"){print $2}}'`"
     #else if ($4 != "Dbase" && $4 != "" && $1 != "Stripe"){print $4}}'`"
     if [ "$FIELDS" != "" ]
     then
     echo "The mirror metadevice on $THISHOST has a problem!"|/usr/ucb/mail
     -s "**** MIRROR FAILIURE !! ****" $WHOCARES
     fi
     done
     fi
     # sleep $HALFHOUR
     #done
     
     
     
     
     The following script is from John Speno:
     
     I wrote a small script which runs periodically from cron. It compares
     the current output of metastat with a master copy which gets created
     when everything is working.
     
     If there are differences (checked using plain old diff), it sends me
     e-mail.
     
     Here's it:
     
     #!/bin/sh
     
     PATH=/usr/bin:/usr/opt/SUNWmd/sbin
     
     STATE=/usr/local/lib/metacheck/master.state
     NEW=/usr/local/lib/metacheck/latest.state
     
     # This defines who gets e-mail when there is a problem
     ADMIN=your-email-address-goes-here
     
     # see if metacheck has already checked, # and no one has fixed the
     problems yet. if [ -f $NEW ]
     then
     exit 0
     fi
     
     # do we have a state file to compare to? if [ ! -f $STATE ]
     then
     echo "Can't read $STATE, giving up"
     exit 1
     fi
     
     # check for differences in the metastat output metastat > $NEW
     
     if [ ! -f $NEW ]
     then
     echo "Can't read $NEW, did metastat run?" exit 1
     fi
     
     diff $STATE $NEW >/dev/null 2>&1
     
     # if things have changed, send a warning if [ $? -eq 1 ]
     then
     mailx -s "Metacheck detected changes" $ADMIN < $NEW
     else
     rm $NEW
     fi
     
     exit 0
     
     
     
______________________________ Reply Separator _________________________________
Subject: Notification of DiskSuite 4.1 error message
Author: Mark_Conroy@em.fcnbd.com at Internet
Date: 7/28/98 7:47 AM
     
     
     Question for the group,
     
     We have been using Sun Enterprise Volume Manager on several of our
     machine and we like the way that we receive notification when a disk
     fails(currently, we have a .forward file set up so that any error
     message sent by Volume Manager, that sends a mail to root, it is
     forwarded to our individual cc-mail addresses.
     
     We have installed DiskSuite 4.1 on a couple of machines, and we
     receive no notification. I went to the manual(novel idea), and there
     is a section identifying that we can have snmp traps sent to a
     management station or there was a reference to setting up a script to
     have messages sent to root's mail. I would prefer to have it go to
     root, and was wondering if anyone had already set up any scripts, for
     notification of when a disk fails, or automatic switch over to hot
     spares has occurred, and would be willing to share.
     
     Just a thought.
     
     Thanks in advance.
     
     Mark Conroy
     
     Will summerize....



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:12:45 CDT