b:11171

[fs/lustre-release.git] / lustre-iokit / obdfilter-survey / obdfilter-survey

diff --git a/lustre-iokit/obdfilter-survey/obdfilter-survey b/lustre-iokit/obdfilter-survey/obdfilter-survey
index 2d85b8f..fd8407d 100755 (executable)
--- a/lustre-iokit/obdfilter-survey/obdfilter-survey
+++ b/lustre-iokit/obdfilter-survey/obdfilter-survey
@@ -38,6 +38,36 @@
 # include library 
 source libecho
 
+# The following variables can be set in the environment, or on the
+# command line
+# result file prefix (date/time + hostname makes unique)
+# NB ensure path to it exists
+rslt_loc=${rslt_loc:-"/tmp"}
+rslt=${rslt:-"$rslt_loc/obdfilter_survey_`date +%F@%R`_`uname -n`"}
+
+# Set this true to check file contents
+verify=${verify:-0}
+
+# total size (MBytes) per obd instance
+# large enough to avoid cache effects 
+# and to make test startup/shutdown overhead insignificant
+size=${size:-16384}
+
+# record size (KBytes)
+rszlo=${rszlo:-1024}
+rszhi=${rszhi:-1024}
+
+# number of objects per OST
+nobjlo=${nobjlo:-1}
+#was nobjhi=${nobjhi:-512}
+nobjhi=${nobjhi:-16}
+
+# threads per OST (1024 max)
+thrlo=${thrlo:-1}
+thrhi=${thrhi:-16}
+
+# End of variables
+
 # create a set of objects, check there are 'n' contiguous ones and
 # return the first or 'ERROR'
 # parameter: 1. hostname
@@ -99,7 +129,7 @@ destroy_objects () {
 
 get_stats () {
     local rfile=$1
-    awk < $rfile \
+    gawk < $rfile \
        '/^Selected device [0-9]+$/ {n = 0; next}\
        /error/ {n = -1; exit}\
        /^[0-9]+\/[0-9]+ Total: [0-9]+\.[0-9]+\/second$/ {n++; v=strtonum($3); \
@@ -173,9 +203,6 @@ if [ -n "$ECHO_CLIENTS" ]; then
     done
 fi
 
-# result file prefix (date/time + hostname makes unique)
-# NB ensure path to it exists
-rslt=${rslt:-"/tmp/obdfilter_survey_`date +%F@%R`_`uname -n`"}
 
 # what tests to run (first must be write)
 tests_str=${tests_str:-""}
@@ -191,27 +218,6 @@ else
     tests=(write rewrite read)
 fi
 
-# Set this true to check file contents
-verify=0
-
-# total size (MBytes) per obd instance
-# large enough to avoid cache effects 
-# and to make test startup/shutdown overhead insignificant
-size=${size:-16384}
-
-# record size (KBytes)
-rszlo=${rszlo:-1024}
-rszhi=${rszhi:-1024}
-
-# number of objects per OST
-nobjlo=${nobjlo:-1}
-#was nobjhi=${nobjhi:-512}
-nobjhi=${nobjhi:-16}
-
-# threads per OST (1024 max)
-thrlo=${thrlo:-1}
-thrhi=${thrhi:-16}
-
 # restart from here iff all are defined
 restart_rsz=
 restart_thr=1
@@ -234,6 +240,10 @@ snap=1
 #####################################################################
 # leave the rest of this alone unless you know what you're doing...
 
+# check and insert obdecho module
+if ! lsmod | grep obdecho > /dev/null; then
+    modprobe obdecho
+fi
 if [ ${#tests[@]} -eq 0 -o "${tests[0]}" != "write" ]; then
     echo "tests: ${tests[@]}"
     echo "First test must be 'write'" 1>&2
@@ -263,34 +273,24 @@ if [ $case == "network" ]; then
        echo "Specify the server NID"
        exit 1;
     fi
-    osc_names_string=`ssh root@"$server_nid" lctl dl`
-    count=0;
-    for name in $osc_names_str; do
-        count=$((count+1))
-    done
-
-    if [ $count != 0 ]; then
-        echo "The existing setup must be cleaned";
-        exit 0;
+    # check for obdecho module on server
+    if ! dsh $server_nid root "lsmod | grep obdecho > /dev/null"; then
+        dsh $server_nid root "modprobe obdecho"
     fi
     # Now do the server setup
-    setup_srv_obd $server_nid "ost_testfs"
-    op_string=`ssh root@"$server_nid" lctl dl`
-    
-    obdecho=0
-    ost=0 
-    for name in $op_string; do
-       if [ "$name" = "obdecho" ]; then
-           obdecho=1
-       fi
-       if [ "$name" = "ost" ]; then
-           ost=1
-       fi
-    done
-
-    if (( $obdecho == 0 || $ost == 0 )); then
-       echo "Server setup not done properly"
-       exit 1
+    setup_srv_obd $server_nid "echo_srv"
+    oss_on_srv=`dsh $server_nid root "lctl dl | grep OSS | awk '{ print $4 }'"`
+    if [ ! $oss_on_srv ]; then
+        setup_OSS $server_nid
+        clean_srv_OSS=1
+    fi
+    if ! dsh $server_nid root "lctl dl | grep obdecho > /dev/null 2>&1"; then
+        echo "obdecho not setup on server"
+        exit 1
+    fi
+    if ! dsh $server_nid root "lctl dl | grep ost > /dev/null 2>&1"; then
+        echo "ost not setup on server"
+        exit 1
     fi
     # Now start client setup
     osc_names_str=$(lctl dl)
@@ -398,6 +398,7 @@ for ((i=0; i < ndevs; i++)); do
     client_names[$i]=${devno[1]}
     do_teardown_ec[$i]=${devno[2]}
 done
+print_summary "$(date) Obdfilter-survey for case=$case from $(hostname)"
 for ((rsz = $rszlo; rsz <= $rszhi; rsz*=2)); do
     for ((nobj = $nobjlo; nobj <= $nobjhi; nobj*=2)); do 
        for ((thr = $thrlo; thr <= $thrhi; thr*=2)); do
@@ -481,9 +482,7 @@ for ((rsz = $rszlo; rsz <= $rszhi; rsz*=2)); do
                 t0=`date +%s.%N`
                 pidcount=0
                 for host in ${unique_hosts[@]}; do
-                   # brutal hack to deal with a non-shared /tmp
-                   scp -q ${cmdsf}_${host} ${host}:/tmp > /dev/null
-                    remote_shell $host bash ${cmdsf}_${host} &
+                    remote_shell $host bash < ${cmdsf}_${host} &
                     pidarray[$pidcount]=$!
                     pidcount=$((pidcount+1))
                 done
@@ -509,8 +508,7 @@ for ((rsz = $rszlo; rsz <= $rszhi; rsz*=2)); do
                    tmpfi="${tmpf}_$idx"
                    echo "=============> $test $client_name" >> $workf
                    host="${host_names[$idx]}"
-                   scp -q ${host}:$tmpfi $tmpfi > /dev/null
-                   cat $tmpfi >> $workf
+                   remote_shell $host cat $tmpfi >> $workf
                    get_stats $tmpfi >> $tmpf
                    rm $tmpfi
                done
@@ -547,6 +545,5 @@ for ((rsz = $rszlo; rsz <= $rszhi; rsz*=2)); do
        done
     done
 done
-
-cleanup 0
+cleanup 0 $clean_srv_OSS
 exit 0