statistics/rpm_dist.r

   1 #####
   2
   3 #system("URL='https://monitor.planet-lab.org:443/monitor/query?object=nodes&nodehistory_hostname=&hostname=on&observed_status=on&rpms=on&rpmvalue=planetlab&tg_format=plain'; curl -s --insecure $URL | grep -v DOWN | grep -v DEBUG | /usr/share/monitor/statistics/hn2rpms.py > out_rpm.csv");
   4 #system("grep MD5SUMS /usr/share/monitor/monitor.log | grep -v measurement-lab | awk 'BEGIN { printf \"hostname,yumsum\\n\" } {if ( $3 != \"\") { printf \"%s,%s\\n\", $2,$3 } }' > yumsum.csv")
   5
   6 r <- read.csv("out_rpm.csv")
   7 ys<- read.csv('yumsum.csv')
   8 m<-merge(r,ys, by="hostname")
   9
  10 s<-table(factor(r$NodeManager), factor(r$kernel), factor(r$iptables));
  11 plot(s);
  12
  13 ideal<-c(NodeManager='NodeManager-1.8-12.planetlab.1',
  14          NodeUpdate='NodeUpdate-0.5-4.planetlab',
  15                  codemux='codemux-0.1-13.planetlab',
  16          fprobe.ulog='fprobe-ulog-1.1.3-0.planetlab',
  17                  ipod='ipod-2.2-1.planetlab',
  18          iproute='iproute-2.6.16-2.planetlab',
  19                  iptables='iptables-1.3.8-9.planetlab',
  20          kernel='kernel-2.6.22.19-vs2.3.0.34.39.planetlab',
  21          madwifi='madwifi-0.9.4-2.6.22.19.3.planetlab',
  22                  monitor.client='monitor-client-3.0-17.planetlab',
  23          monitor.runlevelagent='monitor-runlevelagent-3.0-17.planetlab',
  24                  pl_mom='pl_mom-2.3-1.planetlab',
  25          pl_sshd='pl_sshd-1.0-11.planetlab',
  26                  pyplnet='pyplnet-4.3-3.planetlab',
  27          util.vserver.pl='util-vserver-pl-0.3-17.planetlab',
  28          vserver.planetlab.f8.i386='vserver-planetlab-f8-i386-4.2-12.2009.06.23',
  29          vserver.systemslices.planetlab.f8.i386='vserver-systemslices-planetlab-f8-i386-4.2-12.2009.06.23',
  30          vsys='vsys-0.9-3.planetlab',
  31                  vsys.scripts='vsys-scripts-0.95-11.planetlab');
  32
  33 r_summary <- lapply(r[,4:23], summary)
  34 for (i in 1:length(r_summary))
  35 {
  36     n<-sort(unlist(r_summary[i]), decreasing=TRUE)
  37         names(n[1])
  38 }
  39
  40 as.numeric(factor(ideal[1], levels(r$NodeManager)))
  41
  42 cv <- function ( row , rows=566, start_col=4, end_col=23, ref=NULL)
  43 {
  44         ret<-NULL;
  45     for ( i in 1:rows )
  46         {
  47                 r_l <-NULL
  48             for ( name in names(row) )
  49                 {
  50                         # NOTE: this doesn't work unless the levels in row are a subset of ref's levels.
  51                         x<-as.numeric(factor(row[i,name], levels(factor(unlist(row[name])))));
  52                         r_l <- c(r_l, x);
  53                 }
  54                 #r<-as.numeric(row[i,start_col:end_col]);
  55                 str<- paste(as.character(r_l), collapse="-", sep="-");
  56                 ret<- rbind(ret, str);
  57         }
  58         return (ret);
  59 }
  60
  61 grow <- function (d, column, val)
  62 {
  63     r <- which(d[column] == val);
  64         return (d[r,]);
  65 }
  66
  67 cv(m, length(m$hostname));
  68 i<-data.frame(t(ideal));
  69 cv(i, 1, 1, length(ideal));
  70
  71         # ---
  72
  73 x<-cv(r, length(r$hostname))
  74 x2<-factor(x)
  75 # plot the frequency of each RPM package combination
  76 barplot(sort(table(x2), decreasing=TRUE),
  77                 ylim=c(0, max(table(x2))),
  78                 xlab="Unique Package Combinations",
  79                 ylab="Frequency",
  80                 axisnames=FALSE,
  81                 main=paste("Distribution of Packages for", length(r$hostname),"nodes"));
  82
  83 png("/Users/soltesz/Downloads/rpm_plpackages_distribution_1.png",
  84         width=640,
  85         height=300,
  86         unit="px")
  87 # 1x1 grid, with 1" margins on the bottom/left, 0.5" on the top/right
  88 par(mfrow=c(1,1));
  89 par(mai=c(1,1,0.5,0.5));
  90 barplot(sort(table(x2), decreasing=TRUE),
  91                 ylim=c(0, max(table(x2))),
  92                 xlab="Unique Package Combinations",
  93                 ylab="Frequency",
  94                 axisnames=FALSE,
  95                 main=paste("Distribution of Packages for", length(r$hostname),"nodes"));
  96 dev.off()
  97
  98
  99
 100 #convert_rpm <- function ( row )
 101 #{
 102 #       c <- as.character(row$rpms)
 103 #       rpm_list <- unlist(strsplit(c, " "))
 104 #       rpm_sort <- paste(sort(rpm_list), collapse="::");
 105 #       return (rpm_sort);
 106 #}
 107
 108 #s<-convert_rpm(r)
 109
 110 #for ( row in r[,] )
 111 #{
 112 #       c <- as.character(row$rpms)
 113 #       rpm_list <- unlist(strsplit(c, " "))
 114 #       row$rpm_sort <- paste(sort(rpm_list), collapse="::");
 115 #
 116 #       #for ( rpm in rpm_list )
 117 #       #{
 118 #       #       fields <- unlist(strsplit(rpm, "-"));
 119 #       #       s <- sort(fields);
 120 #       #}
 121 #}
 122 #
 123 #s<-sort(rpm_list);
 124
 125