samba-mirror/ctdb/tools/ctdb_diagnostics

#!/bin/sh
# a script to test the basic setup of a CTDB/Samba install 
# tridge@samba.org September 2007

PATH="$PATH:/sbin:/usr/sbin:/usr/lpp/mmfs/bin"

CONFIG_FILES="/etc/krb5.conf /etc/hosts /etc/ctdb/nodes /etc/sysconfig/ctdb /etc/ctdb/public_addresses /etc/resolv.conf /etc/nsswitch.conf /etc/sysctl.conf /etc/samba/smb.conf /etc/fstab"

2>&1

cat <<EOF
--------------------------------------------------------------------
ctdb_diagnostics starting. This script will gather information about
your ctdb cluster. You should send the output of this script along
with any ctdb or clustered Samba bug reports.
--------------------------------------------------------------------
EOF

date

error() {
    msg="$1"
    echo "ERROR: $msg"
    NUM_ERRORS=`expr $NUM_ERRORS + 1`
    echo " ERROR[$NUM_ERRORS]: $msg" >> $ERRORS
}

show_file() {
    fname="$1"
    echo "  ================================"
    echo "  File: $fname"
    echo "  `ls -l $fname`"
    cat "$fname" | sed 's/^/  /'
    echo "  ================================"
}

show_all() {
    echo "running $1 on all nodes"
    onnode all "hostname; date; $1 2>&1 | sed 's/^/  /'"
}

ERRORS="/tmp/diag_err.$$"
NUM_NODES=`wc -l < /etc/ctdb/nodes`
MAX_NODE=`expr $NUM_NODES - 1`
NUM_ERRORS=0
cat <<EOF
Diagnosis started on a $NUM_NODES node cluster. The following node list will be used:
EOF
show_file /etc/ctdb/nodes


cat <<EOF
--------------------------------------------------------------------
Comping critical config files on all nodes
EOF

for f in $CONFIG_FILES; do
 [ -r "$f" ] || {
    error "$f is missing"
    next;
 }
 show_file $f
 for i in `seq 0 $MAX_NODE`; do
     echo "Testing for same config file $f on node $i"
     tmpf=/tmp/`basename $f`.node$i
     onnode $i cat $f > $tmpf
     cmp $f $tmpf || {
	 error "File $f is different on node $i"
	 diff -u $f $tmpf
     }
     rm -f $tmpf
 done
done

cat <<EOF
--------------------------------------------------------------------
Checking for clock drift
EOF
t=`date +%s`
for i in `seq 0 $MAX_NODE`; do
    t2=`onnode $i date +%s`
    d=`expr $t2 - $t`
    if [ $d -gt 30 -o $d -lt -30 ]; then
	error "time on node $i differs by $d seconds"
    fi
done

cat <<EOF
--------------------------------------------------------------------
Showing software versions
EOF
show_all "uname -a"
[ -x /bin/rpm ] && {
    show_all "rpm -qa | egrep 'samba|ctdb|gpfs'"
}


cat <<EOF
--------------------------------------------------------------------
Showing ctdb status and recent log entries
EOF
show_all "ctdb status; ctdb ip"
show_all "ctdb statistics"

echo "Showing log.ctdb"
show_all "tail -100 /var/log/log.ctdb"

echo "Showing log.ctdb"
show_all "tail -100 /var/log/log.ctdb"


cat <<EOF
--------------------------------------------------------------------
Showing system and process status
EOF
show_all "df; df -i; mount"
show_all uptime
show_all "ps axfw"
show_all "dmesg"
show_all "/sbin/lspci"
show_all "/sbin/ifconfig -a"
show_all "/sbin/ip addr list"
show_all "/sbin/route -n"
show_all "crontab -l"
show_all "sysctl -a"
show_all "/sbin/iptables -L -n"

[ -d /usr/lpp/mmfs ] && {
cat <<EOF
--------------------------------------------------------------------
Showing GPFS status and recent log entries
EOF
 echo "Showing mmfs.log.latest"
 show_all "tail -100 /var/adm/ras/mmfs.log.latest"
 show_all "/usr/lpp/mmfs/bin/mmlsconfig"
 show_all "/usr/lpp/mmfs/bin/mmlsfs all"
 show_all "/usr/lpp/mmfs/bin/mmlsnsd"
 show_all "/usr/lpp/mmfs/bin/mmfsadm dump version"
 show_all "/usr/lpp/mmfs/bin/mmfsadm dump waiters"
 show_all "/usr/lpp/mmfs/bin/mmlsmount all"
 show_all "/usr/lpp/mmfs/bin/mmlsquota"
 show_all "/usr/lpp/mmfs/bin/mmlscluster"
 show_all "/usr/lpp/mmfs/bin/mmlsmgr"
 fslist=`mount|grep type.gpfs|awk '{print $1}'`
 for fs in $fslist; do
     show_all "/usr/lpp/mmfs/bin/mmlssnapshot $fs"
     show_all "/usr/lpp/mmfs/bin/mmlsdisk $fs"
     show_all "/usr/lpp/mmfs/bin/mmlsfileset $fs"
 done
}

cat <<EOF
--------------------------------------------------------------------
Showing Samba status
EOF
show_all "smbstatus -n -B"
show_all "net ads testjoin"
show_all "lsof -n | grep smbd"
show_all "netstat -tan"
show_all "net ads info"
show_all "date"
show_all "smbclient -U% -L 127.0.0.1"
WORKGROUP=`testparm -s --parameter-name=WORKGROUP 2> /dev/null`
show_all id "$WORKGROUP/Administrator"
show_all "wbinfo -p"
show_all "wbinfo --sequence"
show_all "smbd -b"

date
echo "Diagnostics finished with $NUM_ERRORS errors"

[ -r $ERRORS ] && {
    cat $ERRORS
    rm -f $ERRORS
}
exit $NUM_ERRORS