Minor changes to logging so that newsyslog can rotate the logs.

doc/UPDATING

@@ -32,7 +32,11 @@ earliest time one of the steps needs to occur.
 20091228: After install:
 
 	  New daemon to better manage the transition of nodes into
-          hwdown after a failure. Start this daemon:
+          hwdown after a failure. In /etc/newsyslog.conf, add this line:
+
+         /usr/testbed/log/checknodes.log 640 14 300 * Z /var/run/checknodes.pid
+
+          Then start this daemon:
 
 	     boss> sudo /usr/testbed/sbin/checknodes_daemon
 

install/boss-install.in

@@ -737,7 +737,6 @@ Phase "syslog", "Setting up syslog", sub {
 	    "!capserver", "*.*\t\t\t\t\t\t$LOGDIR/capserver.log",
 	    "!frisbeed",  "*.*\t\t\t\t\t\t$LOGDIR/frisbeed.log",
 	    "!pubsubd",    "*.*\t\t\t\t\t\t$LOGDIR/pubsubd.log",
-	    "!stated",    "*.*\t\t\t\t\t\t$LOGDIR/stated.log",
 	    "!osselect",  "*.*\t\t\t\t\t\t$LOGDIR/osselect.log",
 	    "!genlastlog","*.*\t\t\t\t\t\t$LOGDIR/genlastlog.log",
 	    "!sdcollectd","*.*\t\t\t\t\t\t$LOGDIR/sdcollectd.log",
@@ -780,6 +779,8 @@ Phase "syslog", "Setting up syslog", sub {
 	    "$LOGDIR/tmcd.log               640  9     1000 *     Z",
 	    "$LOGDIR/stated.log             640  9     1000 *     Z ".
 	        "/var/run/stated.pid 31",
+	    "$LOGDIR/checknodes.log         640  14    300  *     Z ".
+	        "/var/run/checknodes.pid",
 	    "$LOGDIR/osselect.log           640  9     300  *     Z",
 	    "$LOGDIR/power.log              640  7     300  *     Z",
 	    "$LOGDIR/frisbeed.log           640  7     300  *     Z",

tbsetup/checknodes_daemon.in

 #!/usr/bin/perl -w
 #
 # GENIPUBLIC-COPYRIGHT
-# Copyright (c) 2009 University of Utah and the Flux Group.
+# Copyright (c) 2009-2010 University of Utah and the Flux Group.
 # All rights reserved.
 #
 use strict;
@@ -26,6 +26,7 @@ my $TB		  = "@prefix@";
 my $TBOPS         = "@TBOPSEMAIL@";
 my $TBLOGS        = "@TBLOGSEMAIL@";
 my $LOGFILE       = "$TB/log/checknodes.log";
+my $PIDFILE       = "/var/run/checknodes.pid";
 my $SUDO          = "/usr/local/bin/sudo";
 my $PROTOUSER     = "elabman";
 my $WAP           = "$TB/sbin/withadminprivs";
@@ -76,6 +77,19 @@ use User;
 my $NODEILL_PID = NODEILL_PID();
 my $NODEILL_EID = NODEILL_EID();
 
+#
+# Check for existing pidfile and running daemon. Mostly cause
+# of devel tree versions.
+#
+if (-e $PIDFILE) {
+    my $opid = `cat $PIDFILE`;
+    chomp($opid);
+    if (kill(0, $opid)) {
+	fatal("Lockfile $PIDFILE exists, and process $opid is running");
+    }
+    unlink($PIDFILE);
+}
+
 #
 # We need this user for running below.
 #
@@ -122,6 +136,11 @@ if (! $debug) {
 	exit(0);
     }
 }
+# This acts as a lockfile to prevent a devel version from running at
+# the same time as the real version or another devel version.
+if (system("echo '$PID' > $PIDFILE")) {
+    fatal("Could not create $PIDFILE!");
+}
 print "Check Nodes Daemon starting ... pid $$, at ".`date`;
 
 if ($elabman->FlipTo($experiment->unix_gid())) {
@@ -331,3 +350,4 @@ sub fatal($)
     die("*** $0:\n".
 	"    $msg\n");
 }
+