statistics/node_history_all.r

   1 source("functions.r");
   2
   3 # system("./extract_all.py 2007-* > ../findbad_raw_2007.csv")
   4 # system("./extract_all.py 2008-* > ../findbad_raw_2008.csv")
   5 # system("./extract_all.py 2009-* > ../findbad_raw_2009.csv")
   6
   7 fb7 <- read.csv('findbad_raw_2007.csv', sep=',', header=TRUE)
   8 fb8 <- read.csv('findbad_raw_2008.csv', sep=',', header=TRUE)
   9 fb9 <- read.csv('findbad_raw_2009.csv', sep=',', header=TRUE)
  10
  11 online_nodes <- function (fb)
  12 {
  13     breaks <- unique(fb$timestamp)
  14     n<-NULL
  15     o<-NULL
  16     x<-NULL
  17     for (i in seq(1,length(breaks)) )
  18     {
  19         ts <- breaks[i]
  20         sub <- fb[which(fb$timestamp == ts),]
  21         node_count   <- length(unique(sub$hostname))
  22         online_count <- length(unique(sub$hostname[which(sub$state=='BOOT')]))
  23         x<-c(x,ts)
  24         n<-c(n,node_count)
  25         o<-c(o,online_count)
  26     }
  27     print(length(x))
  28     print(length(n))
  29     print(length(o))
  30     return (rbind(x,n,o))
  31 }
  32 z7<- online_nodes(fb7)
  33 z8<- online_nodes(fb8)
  34 z9<- online_nodes(fb9)
  35
  36 plot(c(z7[1,],z8[1,],z9[1,]), log(c(z7[2,], z8[2,],z9[2,])),
  37         ylim=c(0,7), xlim=c(min(x1), max(x1)), type='p', pch='.', axes=F)
  38 points(c(z7[1,],z8[1,],z9[1,]) , log(c(z7[3,], z8[3,],z9[3,])), pch='.')
  39
  40
  41
  42 t_july08 <-unclass(as.POSIXct("2008-07-01", origin="1970-01-01"))[1]
  43 breaks <- unique(fb8$timestamp[which(fb8$timestamp < t_july08)])
  44 fb8_boot <- fb8$timestamp[which(fb8$state=="BOOT" & fb8$timestamp < t_july08)]
  45 h8<-hist(fb8_boot, breaks=breaks[which(!is.na(breaks) & breaks!=0)])
  46
  47 breaks <- unique(as.numeric(as.character(fb9$timestamp)))
  48 fb9_boot <- as.numeric(as.character(fb9$timestamp[which(fb9$state=="BOOT")]))
  49 hist(fb9_boot, breaks=breaks[which(!is.na(breaks) & breaks >= 1230775020)])
  50