Rev 410: merged from ronnie in http://samba.org/~tridge/ctdb

[tridge at samba.org]

------------------------------------------------------------
revno: 410
revision-id: tridge at samba.org-20070601032911-4o9qwds8gfregd3p
parent: tridge at samba.org-20070531035053-dgk3sbz0t10c7v18
parent: sahlberg at ronnie-20070601032614-3ngwkhvkku56xk27
committer: Andrew Tridgell <tridge at samba.org>
branch nick: tridge
timestamp: Fri 2007-06-01 13:29:11 +1000
message:
  merged from ronnie
modified:
  tools/events                   events-20070529030121-04fjh63cxfh8v1pj-1
  tools/statd-callout            statdcallout-20070531010857-6sdlz455vusye5y5-1
    ------------------------------------------------------------
    revno: 393.1.3
    merged: sahlberg at ronnie-20070601032614-3ngwkhvkku56xk27
    parent: sahlberg at ronnie-20070601031405-yd5dk72l9rtxciz2
    committer: Ronnie Sahlberg <sahlberg at ronnie>
    branch nick: ctdb
    timestamp: Fri 2007-06-01 13:26:14 +1000
    message:
      it is -f   not -x to check if a file exists
    ------------------------------------------------------------
    revno: 393.1.2
    merged: sahlberg at ronnie-20070601031405-yd5dk72l9rtxciz2
    parent: sahlberg at ronnie-20070601030353-6vwmyp86x7rajbh3
    committer: Ronnie Sahlberg <sahlberg at ronnie>
    branch nick: ctdb
    timestamp: Fri 2007-06-01 13:14:05 +1000
    message:
      - create /etc/ctdb/taken_ips and /etc/ctdb/changed_ips analog to the 
      existing /etc/ctdb/released_ips
      
      - only call the statd-callout script if the ips have changed  and call 
      it with a "notify" argument.    we need to restart nfslock service in 
      both cases
      
      - change statd-callout to explicitely restart the lock manager and statd 
      when "notify" is called.   copy the state directory for each held ip 
      from shared storage to /tmp then use sm-notify to send notifications to 
      all monitored clients
    ------------------------------------------------------------
    revno: 393.1.1
    merged: sahlberg at ronnie-20070601030353-6vwmyp86x7rajbh3
    parent: tridge at samba.org-20070530002116-kz22u8npsxt6hhom
    parent: tridge at samba.org-20070531035053-dgk3sbz0t10c7v18
    committer: Ronnie Sahlberg <sahlberg at ronnie>
    branch nick: ctdb
    timestamp: Fri 2007-06-01 13:03:53 +1000
    message:
      new branch from tridges tree
=== modified file 'tools/events'
--- a/tools/events	2007-05-31 01:09:45 +0000
+++ b/tools/events	2007-06-01 03:26:14 +0000
@@ -41,6 +41,9 @@
 		 echo "Failed to add $ip/$maskbits on dev $iface"
 		 exit 1
 	}
+	echo $ip >> /etc/ctdb/taken_ips
+	echo $ip >> /etc/ctdb/changed_ips
+
 	# if we have a local arp entry for this IP then remove it
 	/sbin/arp -d $ip 2> /dev/null
 
@@ -65,19 +68,27 @@
 	# if we have a local arp entry for this IP then remove it
 	/sbin/arp -d $ip 2> /dev/null
 	echo $ip >> /etc/ctdb/released_ips
+	echo $ip >> /etc/ctdb/changed_ips
 	/bin/rm -f /etc/ctdb/ip.$ip
 	exit 0
 	;;
 
      recovered)
-        # restart any services as necessary, like NFS
-	# 
-	[ -x /etc/ctdb/statd-callout ] && /etc/ctdb/statd-callout copy
+	# if we have taken or released any ips we must send out
+	# statd notifications to recover lost nfs locks
+	[ -x /etc/ctdb/statd-callout ] && [ -f /etc/ctdb/changed_ips ] && {
+		/etc/ctdb/statd-callout notify &
+	} >/dev/null 2>&1
+
+        # restart NFS to ensure that all TCP connections to the released ip
+	# are closed
 	[ -f /etc/ctdb/released_ips ] && {
 		( /sbin/service nfs status > /dev/null 2>&1 && 
                       /sbin/service nfs restart > /dev/null 2>&1 ) &
 	} > /dev/null 2>&1
+	/bin/rm -f /etc/ctdb/changed_ips
 	/bin/rm -f /etc/ctdb/released_ips
+	/bin/rm -f /etc/ctdb/taken_ips
 	exit 0
 	;;
 

=== modified file 'tools/statd-callout'
--- a/tools/statd-callout	2007-05-31 01:14:07 +0000
+++ b/tools/statd-callout	2007-06-01 03:14:05 +0000
@@ -22,23 +22,23 @@
 	    /bin/rm -f $STATD_SHARED_DIRECTORY/$ip/$2
 	done
 	;;
-  copy)
-	restart_needed=0
+  notify)
+	# restart the local lock manager and statd
+	/sbin/service nfslock stop > /dev/null 2>&1 
+	/sbin/service nfslock start > /dev/null 2>&1 
+	# send out notifications to any additional ips we now serve
         for f in `/bin/ls /etc/ctdb/ip.*`; do
 	    fname=`/bin/basename $f`
 	    ip=`echo $fname | cut -d. -f2-`
 	    [ -d $STATD_SHARED_DIRECTORY/$ip ] && {
-		/bin/mv $STATD_SHARED_DIRECTORY/$ip $STATD_SHARED_DIRECTORY/$ip.$$
-		/bin/cp -a $STATD_SHARED_DIRECTORY/$ip.$$/. /var/lib/nfs/statd/sm/
-		/bin/rm -rf $STATD_SHARED_DIRECTORY/$ip.$$
-		restart_needed=1
+		# we must copy to a different directory since rpc.statd gets
+		# "upset" if sm-notify touches the files.
+		/bin/rm -rf /tmp/statd/$ip
+		/bin/mkdir -p /tmp/statd/$ip
+		/bin/cp -apr $STATD_SHARED_DIRECTORY/$ip/* /tmp/statd/$ip
+		/usr/sbin/sm-notify -P /tmp/statd/$ip -v $ip -n
 	    }
 	done
-	# restart lockd if necessary
-	[ $restart_needed -eq 1 ] && {
-		( /sbin/service nfslock status > /dev/null 2>&1 && 
-                      /sbin/service nfslock restart > /dev/null 2>&1 ) &
-	} > /dev/null 2>&1
 	;;
 esac