1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306
|
Description: out-of-the-box compatibility with Debian's SGE
Moreover, the patch established the env variable FSLPARALLEL as a uniform
trigger for cluster-based processing in FSL. If set, jobs are submitted to
SGE, if not, not.
--- a/src/fdt/bedpostx
+++ b/src/fdt/bedpostx
@@ -262,16 +262,21 @@
mkdir -p ${subjdir}.bedpostX/logs/monitor
mkdir -p ${subjdir}.bedpostX/xfms
-mailto=`whoami`@fmrib.ox.ac.uk
+# This setting is already present in fsl_sub and allows for a more flexible,
+# centralized configuration
+#mailto=`whoami`@fmrib.ox.ac.uk
echo Queuing preprocessing stages
-preprocid=`${FSLDIR}/bin/fsl_sub -T 60 -m as -N bpx_preproc -l ${subjdir}.bedpostX/logs ${FSLDIR}/bin/bedpostx_preproc.sh ${subjdir} ${gflag}`
+# don't setup email options, rely on fsl_sub's setting
+#preprocid=`${FSLDIR}/bin/fsl_sub -T 60 -m as -N bpx_preproc -l ${subjdir}.bedpostX/logs ${FSLDIR}/bin/bedpostx_preproc.sh ${subjdir} ${gflag}`
+preprocid=`${FSLDIR}/bin/fsl_sub -T 60 -N bpx_preproc -l ${subjdir}.bedpostX/logs ${FSLDIR}/bin/bedpostx_preproc.sh ${subjdir} ${gflag}`
echo Queuing parallel processing stage
nslices=`${FSLDIR}/bin/fslval ${subjdir}/data dim3`
[ -f ${subjdir}.bedpostX/commands.txt ] && rm ${subjdir}.bedpostX/commands.txt
monitor
-if [ "x$SGE_ROOT" = "x" ]; then
+# Don't be SGE-specific -- rely on FSLPARALLEL
+if [ "x$FSLPARALLEL" = "x" ]; then
${subjdir}.bedpostX/monitor&
fi
@@ -293,12 +298,15 @@
fi
slice=$(($slice + 1))
done
-bedpostid=`${FSLDIR}/bin/fsl_sub -j $preprocid -l ${subjdir}.bedpostX/logs -M $mailto -N bedpostx -t ${subjdir}.bedpostX/commands.txt`
+# don't specify a mailto address, but rely on fsl_sub doing it
+bedpostid=`${FSLDIR}/bin/fsl_sub -j $preprocid -l ${subjdir}.bedpostX/logs -N bedpostx -t ${subjdir}.bedpostX/commands.txt`
echo Queuing post processing stage
-mergeid=`${FSLDIR}/bin/fsl_sub -j $bedpostid -T 60 -m as -N bpx_postproc -l ${subjdir}.bedpostX/logs ${FSLDIR}/bin/bedpostx_postproc.sh $subjdir`
+# don't setup email options, rely on fsl_sub's setting
+mergeid=`${FSLDIR}/bin/fsl_sub -j $bedpostid -T 60 -N bpx_postproc -l ${subjdir}.bedpostX/logs ${FSLDIR}/bin/bedpostx_postproc.sh $subjdir`
-if [ "x$SGE_ROOT" != "x" ]; then
+# Don't be SGE-specific -- rely on FSLPARALLEL
+if [ "x$FSLPARALLEL" != "x" ]; then
echo
echo Type ${subjdir}.bedpostX/monitor to show progress.
echo Type ${subjdir}.bedpostX/cancel to terminate all the queued tasks.
--- a/src/fdt/fdt.tcl
+++ b/src/fdt/fdt.tcl
@@ -858,8 +858,11 @@
probtrackx {
global probtrack env
set errorStr ""
- set FSLPARALLEL 0
- if { [ info exists env(SGE_ROOT) ] && $env(SGE_ROOT) != "" } { set FSLPARALLEL 1 }
+ # there seems to be no need to reset FSLPARALLEL since the desired
+ # configuration is already determined in fslstart.tcl (sourced at the
+ #top of this file) in a way that is not SGE-specific.
+ #set FSLPARALLEL 0
+ #if { [ info exists env(SGE_ROOT) ] && $env(SGE_ROOT) != "" } { set FSLPARALLEL 1 }
if { $probtrack(bedpost_dir) == "" } { set errorStr "You must specify the bedpostX directory!" }
if { $probtrack(mode) == "simple" && $probtrack(usereference_yn) && $probtrack(reference) == "" } { set errorStr "$errorStr You must specify a reference image" }
if { $probtrack(mode) == "seedmask" && $probtrack(reference) == "" } { set errorStr "$errorStr You must specify a mask image" }
@@ -1025,7 +1028,11 @@
puts $script "mv $logfile $copylog"
puts $script "rm ${filebase}_script.sh"
close $script
- exec $FSLDIR/bin/fsl_sub -q long.q ${filebase}_script.sh
+ # Don't specify specific queue name (to leave the logic within
+ # fsl_sub, but specify a long time (20 hours) to achieve the same
+ # goal)
+ #exec $FSLDIR/bin/fsl_sub -q long.q ${filebase}_script.sh
+ exec $FSLDIR/bin/fsl_sub -T 1200 ${filebase}_script.sh
} else {
fdt_monitor_short $w "$FSLDIR/bin/probtrackx2 $flags"
--- a/src/misc_tcl/fslstart.tcl
+++ b/src/misc_tcl/fslstart.tcl
@@ -55,7 +55,12 @@
#Choose load_varian specific OR FSLPARALLEL
set FSLPARALLEL 0
-if { [ info exists env(SGE_ROOT) ] && $env(SGE_ROOT) != "" } { set FSLPARALLEL 1 }
+# Instead of relying on SGE_ROOT let's use FSLPARALLEL itself to determine
+# whether using a cluster is desired. This allows for other cluster engines
+# to be used.
+#if { [ info exists env(SGE_ROOT) ] && $env(SGE_ROOT) != "" } { set FSLPARALLEL 1 }
+if { [ info exists env(FSLPARALLEL) ] && $env(FSLPARALLEL) != "" } { set FSLPARALLEL 1 }
+
# what type of OS?
set OSFLAVOUR unix
--- a/src/possum/possumX
+++ b/src/possum/possumX
@@ -251,7 +251,9 @@
mkdir -p ${subjdir}/diff_proc
mkdir -p ${subjdir}/logs
mkdir -p ${subjdir}/logs/pid_${$}
-mailto=`whoami`@fmrib.ox.ac.uk
+# This setting is already present in fsl_sub and allows for a more flexible,
+# centralized configuration
+#mailto=`whoami`@fmrib.ox.ac.uk
echo Preprocessing stage
echo "${POSSUMDIR}/bin/possum_matrix -p ${subjdir}/pulse -m ${subjdir}/motion --seg=$segs -o ${subjdir}/matrix"
@@ -274,9 +276,10 @@
#tejas-add
if [ ! -z $holdjb ]; then hold="-j $holdjb "; fi
-echo "${FSLDIR}/bin/fsl_sub -T $ntime -l ${subjdir}/logs $arch -M $mailto -N possum -t ${subjdir}/possum.com"
+# don't specify a mailto address, but rely on fsl_sub doing it
+echo "${FSLDIR}/bin/fsl_sub -T $ntime -l ${subjdir}/logs $arch -N possum -t ${subjdir}/possum.com"
unset LD_LIBRARY_PATH DYLD_LIBRARY_PATH DYLD_FALLBACK_LIBRARY_PATH
-possumid=`${FSLDIR}/bin/fsl_sub -T $ntime ${hold}-l ${subjdir}/logs $arch -M $mailto -N possum -t ${subjdir}/possum.com`
+possumid=`${FSLDIR}/bin/fsl_sub -T $ntime ${hold}-l ${subjdir}/logs $arch -N possum -t ${subjdir}/possum.com`
#tejas-end
--- a/src/sgeutils/fsl_sub
+++ b/src/sgeutils/fsl_sub
@@ -71,6 +71,13 @@
# cluster.
###########################################################################
+# Domain mail should be sent to if necessary. The complete address will be
+# `whoami`@${fslcluster_mailto_domain}
+fslcluster_mailto_domain=localhost
+# When should the cluster engine send status email (see qsub -m ... docs)
+# Don't send email by default; can be overidden by FSLCLUSTER_MAILOPTS
+# environment variable
+fslcluster_mailopts="${FSLCLUSTER_MAILOPTS:-n}"
###########################################################################
# The following section determines what to do when fsl_sub is called
@@ -92,6 +99,24 @@
fi
fi
+# define the default 'qsub' implementation to be used for cluster submission
+qsub_cmd="qsub"
+
+# Allow to override the above automatic detection result with FSLPARALLEL
+if [ -n "$FSLPARALLEL" ] ; then
+ # Whenever FSLPARALLEL is set enfore using SGE even if no SGE_ROOT is set
+ # which, for example, is the case on Debian systems running SGE
+ METHOD=SGE
+ # TODO: move cluster engine detection here to be able to support more than
+ # just SGE
+ if [ "$FSLPARALLEL" = "condor" ] ; then
+ # if condor shall be used, simply switch to Condor's qsub emulation
+ qsub_cmd="condor_qsub"
+ fi
+else
+ METHOD=NONE
+fi
+
# stop submitted scripts from submitting jobs themselves
if [ "X$FSLSUBALREADYRUN" = "Xtrue" ] ; then
METHOD=NONE
@@ -116,16 +141,22 @@
map_qname ()
{
- if [ $1 -le 20 ] ; then
- queue=veryshort.q
- elif [ $1 -le 120 ] ; then
- queue=short.q
- elif [ $1 -le 1440 ] ; then
- queue=long.q
- else
- queue=verylong.q
- fi
- queueCmd=" -q $queue "
+ # for Debian we can't do the stuff below, because it would be hard
+ # to determine how particular queues are meant to be used on any given
+ # system. Instead of translating into a queue name we specify proper
+ # resource limits, and let SGE decide what queue matches
+ # (qsub wants the time limit in seconds)
+ queueCmd="$queueCmd -l h_rt=$(echo "$1 * 60" | bc)"
+ #if [ $1 -le 20 ] ; then
+ #queue=veryshort.q
+ #elif [ $1 -le 120 ] ; then
+ #queue=short.q
+ #elif [ $1 -le 1440 ] ; then
+ #queue=long.q
+ #else
+ #queue=verylong.q
+ #fi
+ #queueCmd=" -q $queue "
#echo "Estimated time was $1 mins: queue name is $queue"
}
@@ -151,13 +182,13 @@
$command -q short.q gzip *.img *.hdr
$command -a darwin regscript rawdata outputdir ...
- -T <minutes> Estimated job length in minutes, used to auto-set queue name
- -q <queuename> Possible values for <queuename> are "verylong.q", "long.q"
- and "short.q". See below for details
- Default is "long.q".
+ -T <minutes> Estimated job length in minutes, used to let SGE select
+ an appropriate queue
+ -q <queuename> Possible values for <queuename> can be queried with
+ 'qconf -sql'
-a <arch-name> Architecture [e.g., darwin or lx24-amd64]
-p <job-priority> Lower priority [0:-1024] default = 0
- -M <email-address> Who to email, default = `whoami`@fmrib.ox.ac.uk
+ -M <email-address> Who to email, default = `whoami`@localhost
-j <jid> Place a hold on this task until job jid has completed
-t <filename> Specify a task file of commands to execute in parallel
-N <jobname> Specify jobname as it will appear on queue
@@ -170,24 +201,26 @@
<threads> specifies the number of threads to run
-v Verbose mode.
-Queues:
-
-There are several batch queues configured on the cluster, each with defined CPU
-time limits. All queues, except bigmem.q, have a 8GB memory limit.
-
-veryshort.q:This queue is for jobs which last under 30mins.
-short.q: This queue is for jobs which last up to 4h.
-long.q: This queue is for jobs which last less than 24h. Jobs run with a
- nice value of 10.
-verylong.q: This queue is for jobs which will take longer than 24h CPU time.
- There is one slot per node, and jobs on this queue have a nice value
- of 15.
-bigmem.q: This queue is like the verylong.q but has no memory limits.
-
EOF
exit 1
}
+# This part of the help is not applicable anymore
+#
+#Queues:
+#
+#There are several batch queues configured on the cluster, each with defined CPU
+#time limits. All queues, except bigmem.q, have a 8GB memory limit.
+#
+#veryshort.q:This queue is for jobs which last under 30mins.
+#short.q: This queue is for jobs which last up to 4h.
+#long.q: This queue is for jobs which last less than 24h. Jobs run with a
+# nice value of 10.
+#verylong.q: This queue is for jobs which will take longer than 24h CPU time.
+# There is one slot per node, and jobs on this queue have a nice value
+# of 15.
+#bigmem.q: This queue is like the verylong.q but has no memory limits.
+
nargs=$#
if [ $nargs -eq 0 ] ; then
@@ -222,10 +255,13 @@
# change. It also sets up the basic emailing control.
###########################################################################
-queue=long.q
-queueCmd=" -q long.q "
-mailto=`whoami`@fmrib.ox.ac.uk
-MailOpts="a"
+# SGE should already have a default queue, but allow for overwrite
+queueCmd=""
+if [ "x$FSLCLUSTER_DEFAULT_QUEUE" != "x" ] ; then
+ queueCmd=" -q $FSLCLUSTER_DEFAULT_QUEUE "
+fi
+mailto=`whoami`@${fslcluster_mailto_domain}
+MailOpts=${fslcluster_mailopts}
###########################################################################
@@ -423,6 +459,11 @@
exit -1
fi
+ if [ "x$queue" = "x" ] ; then
+ echo $@
+ echo "No queue specified."
+ exit -1
+ fi
# Get a list of queues configured for this PE and confirm that the queue
# we have submitted to has that PE set up.
qstat -g c -pe $peName >/dev/null 2>&1
@@ -447,9 +488,9 @@
if [ "x$tasks" = "x" ] ; then
if [ $scriptmode -ne 1 ] ; then
- sge_command="qsub -V -cwd -shell n -b y -r y $queueCmd $pe_options -M $mailto -N $JobName -m $MailOpts $LogOpts $sge_arch $sge_hold"
+ sge_command="$qsub_cmd -V -cwd -shell n -b y -r y $queueCmd $pe_options -M $mailto -N $JobName -m $MailOpts $LogOpts $sge_arch $sge_hold"
else
- sge_command="qsub $LogOpts $sge_arch $sge_hold"
+ sge_command="$qsub_cmd $LogOpts $sge_arch $sge_hold"
fi
if [ $verbose -eq 1 ] ; then
echo sge_command: $sge_command >&2
@@ -457,7 +498,7 @@
fi
exec $sge_command $@ | awk '{print $3}'
else
- sge_command="qsub -V -cwd $queueCmd $pe_options -M $mailto -N $JobName -m $MailOpts $LogOpts $sge_arch $sge_hold $sge_tasks"
+ sge_command="$qsub_cmd -V -cwd $queueCmd $pe_options -M $mailto -N $JobName -m $MailOpts $LogOpts $sge_arch $sge_hold $sge_tasks"
if [ $verbose -eq 1 ] ; then
echo sge_command: $sge_command >&2
echo control file: $taskfile >&2
|