[devel] [PATCH 1/1] base: Try again for opensafd stop [#2459]
Internally opensafd creates a mutex during start/stop to avoid parallel
execution. Makes mutex more robust and add a short retry if mutex is
taken.
---
src/nid/opensafd.in | 155 +---
1 file changed, 88 insertions(+), 67 deletions(-)
diff --git a/src/nid/opensafd.in b/src/nid/opensafd.in
index d316967c5..57d374361 100644
--- a/src/nid/opensafd.in
+++ b/src/nid/opensafd.in
@@ -196,19 +196,56 @@ check_transport() {
fi
}
+# Create a mutex for start/stop on the filesystem. Will use trap if available.
+mutex_create() {
+ timeout=10
+ interval=2
+ while [ $timeout -gt 0 ]; do
+ if mkdir "$lockfile_inprogress"; then
+ trap 'rmdir "$lockfile_inprogress"; exit $?' INT TERM
EXIT 2> /dev/null
+ return 0
+ else
+ # lockfile exist, try again until timeout
+ if [ $timeout -eq 10 ]; then # log only one time
+ log_warning_msg "opensafd start/stop in
progress. Waiting for lockfile to be removed"
+ logger -t $osafprog "opensafd start/stop in
progress. Waiting for lockfile to be removed"
+ fi
+ sleep $interval
+ timeout=$((timeout-interval))
+ fi
+ done
+
+ log_warning_msg "opensafd start/stop already in progress. Unable to
continue"
+ logger -t $osafprog "opensafd start/stop already in progress. Unable to
continue"
+ log_warning_msg "To forcefully start/stop OpenSAF remove
$lockfile_inprogress"
+ logger -t $osafprog "To forcefully start/stop OpenSAF remove
$lockfile_inprogress"
+ return 1
+}
+
+mutex_remove() {
+ rmdir "$lockfile_inprogress" 2> /dev/null
+ trap - INT TERM EXIT 2> /dev/null
+}
+
start() {
+ if ! mutex_create; then
+ return 1
+ fi
+
export LD_LIBRARY_PATH=$pkglibdir:$LD_LIBRARY_PATH
pidofproc -p $amfnd_pid $amfnd_bin > /dev/null 2>&1
lsb_status=$?
if [ $lsb_status -eq 0 ]; then
- RETVAL=0
+ RETVAL=0
log_success_msg
+ mutex_remove
return $RETVAL
fi
[ -x $daemon ] || exit 5
+# Does more than check ...
check_env
check_transport
@@ -218,85 +255,69 @@ start() {
#enable_coredump
echo -n "Starting OpenSAF Services (Using $MDS_TRANSPORT):"
- if [ -e "$lockfile_inprogress" ]; then
- RETVAL=1
- log_warning_msg "opensafd start/stop already in progress.
Unable to continue"
- logger -t $osafprog "opensafd start/stop already in progress.
Unable to continue"
- log_warning_msg "To forcefully start/stop OpenSAF remove
$lockfile_inprogress"
- logger -t $osafprog "To forcefully start/stop OpenSAF remove
$lockfile_inprogress"
+ start_daemon $binary $args
+ RETVAL=$?
+ if [ $RETVAL -eq 0 ]; then
+ logger -t $osafprog "OpenSAF($osafversion - $osafcshash)
services successfully started"
+ touch $lockfile
+ log_success_msg
else
- touch "$lockfile_inprogress"
- start_daemon $binary $args
- RETVAL=$?
- if [ $RETVAL -eq 0 ]; then
- logger -t $osafprog "OpenSAF($osafversion -
$osafcshash) services successfully started"
- touch $lockfile
- log_success_msg
+ final_clean
+ log_failure_msg
+ if [ $REBOOT_ON_FAIL_TIMEOUT -ne 0 ]; then
+ logger -t $osafprog "Starting OpenSAF failed,
rebooting..."
+ sleep $REBOOT_ON_FAIL_TIMEOUT
+ mutex_remove
+ /sbin/reboot &
else
- final_clean
- log_failure_msg
- if [ $REBOOT_ON_FAIL_TIMEOUT -ne 0 ]; then
- logger -t $osafprog "Starting OpenSAF failed,
rebooting..."
- sleep $REBOOT_ON_FAIL_TIMEOUT
- rm -f "$lockfile_inprogress"
- /sbin/reboot &
- else
- logger -t $osafprog "Starting OpenSAF failed"
- fi
+ logger -t $osafprog "Starting OpenSAF failed"
fi
- rm -f "$lockfile_inprogress"
fi
+ mutex_remove
return $RETVAL
}
stop() {
- logger -t $osafprog "Stopping OpenSAF Services"
+ if ! mutex_create; then
+ return 1
+ fi
- if [ -e "$lockfile_inprogress" ]; then
- RETVAL=1
- log_warning_msg "opensafd start/stop
Re: [devel] [PATCH 1/1] base: Try again for opensafd stop [#2459]
Note that we maybe should avoid adding new a dependency towards the
inotifywait tool...
regards,
Anders Widell
On 06/02/2017 02:38 PM, Rafael Odzakow wrote:
Thanks I am trying to implement the suggestions and will send out a
new patch soon.
On 06/02/2017 08:51 AM, Hans Nordebäck wrote:
Hi Rafael,
I forgot one thing, not to say that you should change your patch,
only as an example as we use inotify in opensaf,
the following may also work:
wait_for_lockfile_clear() {
inotifywait -q -t 20 -e delete_self $lockfile_inprogress
if [ $? = 2 ]; then
return 1
else
touch $lockfile_inprogress
return 0
fi
}
/Hans
On 06/02/2017 07:27 AM, Hans Nordebäck wrote:
ack, code review only.
Minor comments, in current opensafd script, the test and create of
the lockfile is not atomic, so there is
a window for race. Perhaps we can move the creation of the lockfile
to function "wait_for_lockfile_clear",
and make the test and create of the lockfile atomic?
/HansN
On 06/01/2017 02:37 PM, Rafael Odzakow wrote:
Internally opensafd creates a lock file during start/stop to avoid
parallel execution. Wait for this lockfile to be released when a
call to
opensafd stop is done.
---
src/nid/opensafd.in | 19 +--
1 file changed, 17 insertions(+), 2 deletions(-)
diff --git a/src/nid/opensafd.in b/src/nid/opensafd.in
index e7683bd7e..df90331a6 100644
--- a/src/nid/opensafd.in
+++ b/src/nid/opensafd.in
@@ -196,6 +196,22 @@ check_transport() {
fi
}
+wait_for_lockfile_clear() {
+local timeout=10
+local interval=2
+while [ $timeout -gt 0 ]; do
+if [ -e "$lockfile_inprogress" ]; then
+log_warning_msg "opensafd start/stop in
progress. Wait for lockfile to be removed"
+logger -t $osafprog "opensafd start/stop in
progress. Wait for lockfile to be removed"
+sleep $interval
+else
+return 0
+fi
+timeout=`expr $timeout - $interval`
+done
+return 1
+}
+
start() {
export LD_LIBRARY_PATH=$pkglibdir:$LD_LIBRARY_PATH
pidofproc -p $amfnd_pid $amfnd_bin > /dev/null 2>&1
@@ -251,8 +267,7 @@ start() {
stop() {
logger -t $osafprog "Stopping OpenSAF Services"
-
-if [ -e "$lockfile_inprogress" ]; then
+if ! wait_for_lockfile_clear; then
RETVAL=1
log_warning_msg "opensafd start/stop already in progress.
Unable to continue"
logger -t $osafprog "opensafd start/stop already in
progress. Unable to continue"
--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
___
Opensaf-devel mailing list
Opensaf-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opensaf-devel
Re: [devel] [PATCH 1/1] base: Try again for opensafd stop [#2459]
Thanks I am trying to implement the suggestions and will send out a new
patch soon.
On 06/02/2017 08:51 AM, Hans Nordebäck wrote:
Hi Rafael,
I forgot one thing, not to say that you should change your patch, only
as an example as we use inotify in opensaf,
the following may also work:
wait_for_lockfile_clear() {
inotifywait -q -t 20 -e delete_self $lockfile_inprogress
if [ $? = 2 ]; then
return 1
else
touch $lockfile_inprogress
return 0
fi
}
/Hans
On 06/02/2017 07:27 AM, Hans Nordebäck wrote:
ack, code review only.
Minor comments, in current opensafd script, the test and create of
the lockfile is not atomic, so there is
a window for race. Perhaps we can move the creation of the lockfile
to function "wait_for_lockfile_clear",
and make the test and create of the lockfile atomic?
/HansN
On 06/01/2017 02:37 PM, Rafael Odzakow wrote:
Internally opensafd creates a lock file during start/stop to avoid
parallel execution. Wait for this lockfile to be released when a
call to
opensafd stop is done.
---
src/nid/opensafd.in | 19 +--
1 file changed, 17 insertions(+), 2 deletions(-)
diff --git a/src/nid/opensafd.in b/src/nid/opensafd.in
index e7683bd7e..df90331a6 100644
--- a/src/nid/opensafd.in
+++ b/src/nid/opensafd.in
@@ -196,6 +196,22 @@ check_transport() {
fi
}
+wait_for_lockfile_clear() {
+local timeout=10
+local interval=2
+while [ $timeout -gt 0 ]; do
+if [ -e "$lockfile_inprogress" ]; then
+log_warning_msg "opensafd start/stop in
progress. Wait for lockfile to be removed"
+logger -t $osafprog "opensafd start/stop in
progress. Wait for lockfile to be removed"
+sleep $interval
+else
+return 0
+fi
+timeout=`expr $timeout - $interval`
+done
+return 1
+}
+
start() {
export LD_LIBRARY_PATH=$pkglibdir:$LD_LIBRARY_PATH
pidofproc -p $amfnd_pid $amfnd_bin > /dev/null 2>&1
@@ -251,8 +267,7 @@ start() {
stop() {
logger -t $osafprog "Stopping OpenSAF Services"
-
-if [ -e "$lockfile_inprogress" ]; then
+if ! wait_for_lockfile_clear; then
RETVAL=1
log_warning_msg "opensafd start/stop already in progress.
Unable to continue"
logger -t $osafprog "opensafd start/stop already in
progress. Unable to continue"
--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
___
Opensaf-devel mailing list
Opensaf-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opensaf-devel
[devel] [PATCH 1/1] base: Try again for opensafd stop [#2459]
Internally opensafd creates a lock file during start/stop to avoid
parallel execution. Wait for this lockfile to be released when a call to
opensafd stop is done.
---
src/nid/opensafd.in | 19 +--
1 file changed, 17 insertions(+), 2 deletions(-)
diff --git a/src/nid/opensafd.in b/src/nid/opensafd.in
index e7683bd7e..df90331a6 100644
--- a/src/nid/opensafd.in
+++ b/src/nid/opensafd.in
@@ -196,6 +196,22 @@ check_transport() {
fi
}
+wait_for_lockfile_clear() {
+local timeout=10
+local interval=2
+while [ $timeout -gt 0 ]; do
+if [ -e "$lockfile_inprogress" ]; then
+log_warning_msg "opensafd start/stop in progress. Wait for
lockfile to be removed"
+logger -t $osafprog "opensafd start/stop in progress. Wait
for lockfile to be removed"
+sleep $interval
+else
+return 0
+fi
+timeout=`expr $timeout - $interval`
+done
+return 1
+}
+
start() {
export LD_LIBRARY_PATH=$pkglibdir:$LD_LIBRARY_PATH
pidofproc -p $amfnd_pid $amfnd_bin > /dev/null 2>&1
@@ -251,8 +267,7 @@ start() {
stop() {
logger -t $osafprog "Stopping OpenSAF Services"
-
- if [ -e "$lockfile_inprogress" ]; then
+if ! wait_for_lockfile_clear; then
RETVAL=1
log_warning_msg "opensafd start/stop already in progress.
Unable to continue"
logger -t $osafprog "opensafd start/stop already in progress.
Unable to continue"
--
2.11.0
--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
___
Opensaf-devel mailing list
Opensaf-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opensaf-devel
