various tricks and notes for smoother debugging

diff --git a/Makefile b/Makefile
index eeab636..1a55599 100644 (file)
--- a/Makefile
+++ b/Makefile
@@ -24,6 +24,7 @@ install:
            --install-purelib=$(DESTDIR)/$(datadir)/NodeManager \
            --install-platlib=$(DESTDIR)/$(datadir)/NodeManager \
            --install-scripts=$(DESTDIR)/$(bindir)
+       install -m 444 README $(DESTDIR)/$(datadir)/NodeManager
 
 clean:
        python setup.py clean
@@ -37,3 +38,57 @@ tags:
 
 .PHONY: tags
 
+########## sync
+# for use with the test framework; push local stuff on a test node
+# make sync NODE=vnode01.inria.fr
+# specify TESTMASTER and BUILD if the key is not available yet
+
+LOCAL_RSYNC_EXCLUDES   := --exclude '*.pyc' 
+RSYNC_EXCLUDES         := --exclude .svn --exclude CVS --exclude '*~' --exclude TAGS $(LOCAL_RSYNC_EXCLUDES)
+RSYNC_COND_DRY_RUN     := $(if $(findstring n,$(MAKEFLAGS)),--dry-run,)
+RSYNC                  := rsync -e "ssh -i $(NODE).key.rsa" -a -v $(RSYNC_COND_DRY_RUN) $(RSYNC_EXCLUDES)
+
+ifdef NODE
+NODEURL:=root@$(NODE):/
+endif
+
+sync: $(NODE).key.rsa
+ifeq (,$(NODEURL))
+       @echo "sync: You must define NODE on the command line"
+       @echo "  e.g. make sync NODE=vnode01.inria.fr"
+       @exit 1
+else
+       +$(RSYNC) ./ $(NODEURL)/usr/share/NodeManager/
+       +$(RSYNC) ./initscripts/nm $(NODEURL)/etc/init.d/nm
+endif
+
+### fetching the key
+
+TESTMASTER ?= testmaster.onelab.eu
+
+ifdef BUILD
+KEYURL:=root@$(TESTMASTER):$(BUILD)/keys/key1.rsa
+endif
+
+$(NODE).key.rsa:
+ifeq (,$(KEYURL))
+       @echo "sync: fetching $@ - You must define TESTMASTER, BUILD and NODE on the command line"
+       @echo "  e.g. make sync TESTMASTER=testmaster.onelab.eu BUILD=2010.01.22--1l-f8-32 NODE=vnode01.inria.fr"
+       @echo "  note that for now all test builds use the same key, so any BUILD would do"
+       @exit 1
+else
+       scp $(KEYURL) $@
+endif
+
+### utility - find out the node name for a given BUILD
+
+ifdef BUILD
+NODEIPCOMMAND:=ssh root@$(TESTMASTER) cat $(BUILD)/arg-ips-node
+endif
+
+nodename:
+ifeq (,$(NODEIPCOMMAND))
+       @echo "nodename: You must define TESTMASTER and BUILD on the command line"
+else
+       $(NODEIPCOMMAND)
+endif

diff --git a/NodeManager.spec b/NodeManager.spec
index 978280d..0f96ecd 100644 (file)
--- a/NodeManager.spec
+++ b/NodeManager.spec
@@ -61,9 +61,11 @@ local operations on slices.
 %setup -q
 
 %build
+# make manages the C and Python stuff
 %{__make} %{?_smp_mflags}
 
 %install
+# make manages the C and Python stuff
 rm -rf $RPM_BUILD_ROOT
 %{__make} %{?_smp_mflags} install DESTDIR="$RPM_BUILD_ROOT"
 

diff --git a/README b/README
index d049d21..c6233eb 100644 (file)
--- a/README
+++ b/README
@@ -5,3 +5,32 @@ The PlanetLab NodeManager manages all aspects of PlanetLab node and slice manage
 
 Copyright 2008 Princeton University
 http://www.planet-lab.org
+
+========================================
+--- NOTES on debugging
+
+
+(*) service nm restartdebug
+this lets you restart the nodemanager in interactive mode, and with faster periods
+
+
+(*) logs
+logging is a bit confusing; the main recipient is 
+    /var/log/nm
+together with
+    /var/log/nm.daemon (which should be empty)
+
+but beware that part of the logs eventually end up in 
+    /vservers/<slicename>/var/log/nm
+as well as in here (although this should not be used anymore by now)
+    /vservers/<slicename>/var/log/boot.log
+
+
+(*) connecting with PLC
+also, note that each time GetSlivers gets sucessfully fetched, it is saved in
+    /var/log/getslivers.txt
+
+When the connection is broken, for obscure reasons, all modules are
+invoked anyway, but with an empty dict as data, triggering various
+exceptions that complain about missing keys. In general these are red
+herrings.

diff --git a/initscripts/nm b/initscripts/nm
index 61b7254..375e157 100755 (executable)
--- a/initscripts/nm
+++ b/initscripts/nm
@@ -20,6 +20,8 @@ nm=${NM-"python /usr/share/NodeManager/nm.py"}
 prog="Node Manager"
 options=${OPTIONS-"-d -s"}
 restartoptions=${RESTARTOPTIONS-"-d"}
+# debug mode is interactive, and has faster period
+debugoptions=${DEBUGOPTIONS-"-p 60 -r 31"}
 pidfile=${PIDFILE-/var/run/nm.pid}
 lockfile=${LOCKFILE-/var/lock/subsys/nm}
 RETVAL=0
@@ -48,34 +50,33 @@ stop()
     [ $RETVAL -eq 0 ] && rm -f ${lockfile} ${pidfile}
 }
 
-restart()
-{
-    stop
-    do_start $restartoptions
-}
-
-
 case "$1" in
     start)
-    start
-    ;;
+       start
+       ;;
     stop)
-    stop
-    ;;
+       stop
+       ;;
     status)
-    status $nm
-    RETVAL=$?
-    ;;
+       status $nm
+       RETVAL=$?
+       ;;
     restart|reload)
-    restart
-    ;;
+       stop
+       do_start $restartoptions
+       ;;
+    restartdebug)
+       stop
+       echo "Running interactively .."
+       $nm $debugoptions
+       ;;
     condrestart)
-    if [ -f ${pidfile} ] ; then
-        restart
-    fi
+       if [ -f ${pidfile} ] ; then
+            restart
+       fi
     ;;
     *)
-    echo $"Usage: $0 {start|stop|restart|condrestart|status}"
+    echo $"Usage: $0 {start|stop|restart|condrestart|status|restartdebug}"
     exit 1
 esac
 

diff --git a/nm.py b/nm.py
index 0776826..72fbd80 100755 (executable)
--- a/nm.py
+++ b/nm.py
@@ -35,13 +35,18 @@ known_modules=['net','conf_files', 'sm', 'bwmon']
 
 plugin_path = "/usr/share/NodeManager/plugins"
 
+default_period=600
+default_random=301
+
 parser = optparse.OptionParser()
 parser.add_option('-d', '--daemon', action='store_true', dest='daemon', default=False, help='run daemonized')
 parser.add_option('-s', '--startup', action='store_true', dest='startup', default=False, help='run all sliver startup scripts')
 parser.add_option('-f', '--config', action='store', dest='config', default='/etc/planetlab/plc_config', help='PLC configuration file')
 parser.add_option('-k', '--session', action='store', dest='session', default='/etc/planetlab/session', help='API session key (or file)')
-parser.add_option('-p', '--period', action='store', dest='period', default=600, help='Polling interval (sec)')
-parser.add_option('-r', '--random', action='store', dest='random', default=301, help='Range for additional random polling interval (sec)')
+parser.add_option('-p', '--period', action='store', dest='period', default=default_period, 
+                  help='Polling interval (sec) - default %d'%default_period)
+parser.add_option('-r', '--random', action='store', dest='random', default=default_random, 
+                  help='Range for additional random polling interval (sec) -- default %d'%default_random)
 parser.add_option('-v', '--verbose', action='store_true', dest='verbose', default=False, help='more verbose log')
 parser.add_option('-P', '--path', action='store', dest='path', default=plugin_path, help='Path to plugins directory')