statistics/operator_overhead.r

   1 source("functions.r");
   2
   3 median_time_to_resolve_window <- function (t, from, to, window, fmt="%b")
   4 {
   5     # find 'type' range of days
   6     dates <-seq(as.Date(from), as.Date(to), 'week')
   7     months <- format(dates, fmt)
   8     hbreaks<-unclass(as.POSIXct(dates))
   9
  10     xx<-NULL;
  11     yy<-NULL;
  12     yy_sd_high<-NULL;
  13     yy_sd_low<-NULL;
  14     date_index <- NULL;
  15     q_list <- NULL;
  16
  17     x<-seq(-20,20,0.01)
  18     for ( i in seq(1,length(hbreaks)-window-1) )
  19     {
  20         print (sprintf("round %s of %s", i, length(hbreaks)-window-1))
  21         # get range from t
  22         t_sub <- t[which(t$start > hbreaks[i] & t$start<= hbreaks[i+window]),]
  23         if ( length(t_sub$start) <= 1 )  { next }
  24         # take log, then sn.mle -> h
  25         d <- (t_sub$lastreply - t_sub$start)/(60*60)    # hours
  26         d <- log(d)                                     # log(hours)
  27             # sn.mle
  28         print (sprintf("length: %s", length(d)))
  29         q<-quantile(d)
  30         print(q)
  31
  32         date_index <- c(date_index, round(i+window/2))
  33
  34         xx<- c(xx, hbreaks[round(i+window/2)])
  35         q_list <- rbind(q_list, q)
  36
  37     }
  38     m<- months[date_index]
  39     return (cbind(xx,q_list, m))
  40 }
  41
  42 available_nodes <- function (ns, from, to, type, fmt="%b")
  43 {
  44     # find 'type' range of days
  45     dates <-seq(as.Date(from), as.Date(to), type)
  46     months <- format(dates, fmt)
  47     hbreaks<-unclass(as.POSIXct(dates))
  48
  49     xx<-NULL;
  50     yy<-NULL;
  51
  52     for ( i in seq(1,length(hbreaks)-1) )
  53     {
  54         # get range from ns
  55         ns_sub <- ns[which(ns$date > hbreaks[i] & ns$date <= hbreaks[i+1] & ns$status == 'BOOT'),]
  56         nodes <- length(ns_sub$date)
  57
  58         xx<- c(xx, hbreaks[i])
  59         yy<- c(yy, nodes)
  60
  61     }
  62     m<- months[1:length(months)-1]
  63     return (rbind(xx,yy,m))
  64 }
  65
  66
  67
  68 open_tickets <- function (t, from, to, type, fmt="%b")
  69 {
  70     # find 'type' range of days
  71     dates <-seq(as.Date(from), as.Date(to), type)
  72     months <- format(dates, fmt)
  73     hbreaks<-unclass(as.POSIXct(dates))
  74
  75     xx<-NULL;
  76     yy<-NULL;
  77
  78     for ( i in seq(1,length(hbreaks)-1) )
  79     {
  80         # identify any tickets with a start time in range, lastreply in range
  81         # or where both start is less and lastreply is greater than the range
  82         t_sub <- t[which( (t$start < hbreaks[i] & t$lastreply > hbreaks[i+1]) |
  83                           (t$start > hbreaks[i] & t$start <= hbreaks[i+1]) |
  84                           (t$lastreply > hbreaks[i] & t$lastreply <= hbreaks[i+1]) ),]
  85         tickets <- length(t_sub$start)
  86         #if ( nrow(t_sub) > 0 ){
  87         #    for ( j in seq(1,nrow(t_sub)) )
  88         #    {
  89         #        #print(sprintf("id %s, date %s", t_sub[i,'ticket_id'], t_sub[i,'s1']))
  90         #        print(sprintf("id %s, date %s", t_sub[j,]$ticket_id, t_sub[j, 's1']))
  91         #    }
  92         #}
  93
  94         xx<- c(xx, hbreaks[i])
  95         yy<- c(yy, tickets)
  96
  97     }
  98     m<- months[1:length(months)-1]
  99     return (rbind(xx,yy,m))
 100 }
 101
 102 online_nodes <- function (fb)
 103 {
 104     breaks <- unique(fb$timestamp)
 105     n<-NULL
 106     o<-NULL
 107     x<-NULL
 108     for (i in seq(1,length(breaks)) )
 109     {
 110         ts <- breaks[i]
 111         sub <- fb[which(fb$timestamp == ts),]
 112         node_count   <- length(unique(sub$hostname))
 113         online_count <- length(unique(sub$hostname[which(sub$state=='BOOT')]))
 114         x<-c(x,ts)
 115         n<-c(n,node_count)
 116         o<-c(o,online_count)
 117     }
 118     print(length(x))
 119     print(length(n))
 120     print(length(o))
 121     return (rbind(x,n,o))
 122 }
 123
 124 lowess_smooth <- function (x, y, delta=(60*60*24), f=0.02)
 125 {
 126     a<-lowess(x, y, delta=delta, f=f)
 127     return (a);
 128 }
 129
 130 #####
 131
 132 # system("parse_rt_data.py 3 > rt_data.csv");
 133 t <- read.csv('rt_data_2004-2010.csv', sep=',', header=TRUE)
 134 t2 <- t[which(t$complete == 1),]
 135 ot <- open_tickets(t2, '2004/1/1', '2010/2/28', 'day', "%b")
 136
 137 start_image("rt_operator_overhead.png")
 138 par(mfrow=c(2,1))
 139 par(mai=c(0,1,0.3,0.1))
 140
 141 x1<-as.numeric(ot[1,])
 142 y1<-as.numeric(ot[2,])
 143
 144 a_ot<-lowess_smooth(x1, y1)
 145
 146 plot(x1, y1, col='grey80', type='l', axes=F,
 147     ylab="a) Open Tickets (tickets/day)", xlab="Date",
 148     ylim=c(0,120)) # , ylim=c(0,260))
 149 lines(a_ot$x, round(a_ot$y), col='black')
 150
 151 #axis(1, labels=ot[3,], at=ot[1,], cex.axis=0.7)
 152 axis(2, las=1)
 153 #mtext("2004           2005           2006           2007           2008           2009", 1,2)
 154
 155 #abline_at_date('2005-01-01', 'grey60')
 156 #abline_at_date('2006-01-01', 'grey60')
 157 #abline_at_date('2007-01-01', 'grey60')
 158 #abline_at_date('2008-01-01', 'grey60')
 159 #abline_at_date('2009-01-01', 'grey60')
 160 #abline_at_date('2010-01-01', 'grey60')
 161 abline(h=25, lty=3, col='grey80')
 162 abline(h=40, lty=3, col='grey80')
 163
 164 tstamp_20040412 <-abline_at_date("2004-04-12", col='white', lty=0, height=110)
 165 tstamp_20041112 <-abline_at_date("2004-11-12", col='grey60', lty=3, height=110)
 166 tstamp_20050301 <-abline_at_date("2005-03-01", col='grey60', lty=3, height=110)
 167 tstamp_20050615 <-abline_at_date("2005-06-15", col='white',  lty=0, height=110)
 168 #tstamp_20051023 <-abline_at_date("2005-10-23", col='grey60', lty=3, height=110)
 169 tstamp_20051001 <-abline_at_date("2005-10-01", col='grey60', lty=3, height=110)
 170 tstamp_20060519 <-abline_at_date("2006-05-19", col='grey60', lty=3, height=110)
 171 tstamp_20070228 <-abline_at_date("2007-02-28", col='grey60', lty=3, height=110)
 172 tstamp_20070501 <-abline_at_date("2007-05-01", col='white',  lty=0, height=110)
 173 tstamp_20071021 <-abline_at_date("2007-10-21", col='grey60', lty=3, height=110)
 174 tstamp_20080601 <-abline_at_date("2008-06-01", col='grey60', lty=3, height=110)
 175 tstamp_20080815 <-abline_at_date("2008-08-15", col='white',  lty=0, height=110)
 176 tstamp_20090501 <-abline_at_date("2009-05-01", col='grey60', lty=3, height=110)
 177 tstamp_20100201 <-abline_at_date("2010-02-01", col='white',  lty=0, height=110)
 178
 179         #tstamp_20051023,
 180 text(x=c(tstamp_20040412,
 181         tstamp_20041112,
 182         tstamp_20050301,
 183         tstamp_20050615,
 184         tstamp_20051001,
 185         tstamp_20060519,
 186         tstamp_20070228,
 187         tstamp_20071021,
 188         tstamp_20080601,
 189         tstamp_20090501),
 190      y=c(120),
 191      labels=c('Release', '3.0', '3.1', '', '3.2', '3.3', '4.0', '4.1', '4.2', '4.3'))
 192      #labels=c('3.0', '3.1', '3.1S', '3.2', '4.0', '4.2', '4.3'))
 193
 194
 195 #text(x=c( tstamp_20041112+(tstamp_20050301-tstamp_20041112)/2,
 196 #        tstamp_20050301+(tstamp_20050615-tstamp_20050301)/2,
 197 #        tstamp_20050615+(tstamp_20051023-tstamp_20050615)/2,
 198 #        tstamp_20051023+(tstamp_20070101-tstamp_20051023)/2,
 199 #        tstamp_20070101+(tstamp_20070501-tstamp_20070101)/2,
 200 #        tstamp_20080601+(tstamp_20080815-tstamp_20080601)/2,
 201 #        tstamp_20090501+(tstamp_20100201-tstamp_20090501)/2 ),
 202 #     y=c(120),
 203 #     labels=c('3.0', '3.1', '3.1S', '3.2', '4.0', '4.2', '4.3'))
 204
 205 par(mai=c(1,1,0.1,0.1))
 206 for ( s in c(7) )
 207 {
 208     d<- median_time_to_resolve_window(t2, "2004/1/1", "2010/2/28", s, "%b")
 209     plot(d[,1], exp(as.numeric(d[,5]))/24, type='l', lty=1, xlab="",
 210             axes=F, ylim=c(0.01, 15), ylab="b) Resolution Time by", col='black',
 211             xlim=c(min(x1), max(x1)))
 212     mtext("Quartile (days)", 2, 2)
 213     lines(d[,1], exp(as.numeric(d[,4]))/24, lty=1, col='grey50')
 214     lines(d[,1], exp(as.numeric(d[,3]))/24, lty=1, col='grey75')
 215     #axis(1, labels=d[,7], at=d[,1])
 216     axis(1, labels=ot[3,], at=ot[1,], cex.axis=0.7)
 217     mtext("2004           2005           2006           2007           2008           2009", 1,2)
 218     axis(2, labels=c(0,1,4,7,14), at=c(0,1,4,7,14), las=1)
 219     m<-round(max(exp(as.numeric(d[,4]))/24), 2)
 220     #axis(2, labels=m, at=m, las=1)
 221     #abline(h=m, lty=3, col='grey80')
 222 }
 223
 224 abline(h=1, lty=3, col='grey80')
 225 abline(h=4, lty=3, col='grey80')
 226 abline(h=7, lty=3, col='grey80')
 227
 228 tstamp_20040412 <-abline_at_date("2004-04-12", col='white', lty=0, height=14)
 229 tstamp_20041112 <-abline_at_date("2004-11-12", col='grey60', lty=3, height=14)
 230 tstamp_20050301 <-abline_at_date("2005-03-01", col='grey60', lty=3, height=14)
 231 tstamp_20050615 <-abline_at_date("2005-06-15", col='white',  lty=0, height=14)
 232 #tstamp_20051023 <-abline_at_date("2005-10-23", col='grey60', lty=3, height=14)
 233 tstamp_20051001 <-abline_at_date("2005-10-01", col='grey60', lty=3, height=14)
 234 tstamp_20060519 <-abline_at_date("2006-05-19", col='grey60', lty=3, height=14)
 235 tstamp_20070228 <-abline_at_date("2007-02-28", col='grey60', lty=3, height=14)
 236 tstamp_20070501 <-abline_at_date("2007-05-01", col='white',  lty=0, height=14)
 237 tstamp_20071021 <-abline_at_date("2007-10-21", col='grey60', lty=3, height=14)
 238 tstamp_20080601 <-abline_at_date("2008-06-01", col='grey60', lty=3, height=14)
 239 tstamp_20080815 <-abline_at_date("2008-08-15", col='white',  lty=0, height=14)
 240 tstamp_20090501 <-abline_at_date("2009-05-01", col='grey60', lty=3, height=14)
 241 tstamp_20100201 <-abline_at_date("2010-02-01", col='white',  lty=0, height=14)
 242
 243
 244 text(x=c(tstamp_20040412,
 245         tstamp_20041112,
 246         tstamp_20050301,
 247         tstamp_20050615,
 248         tstamp_20051001,
 249         tstamp_20060519,
 250         tstamp_20070228,
 251         tstamp_20071021,
 252         tstamp_20080601,
 253         tstamp_20090501),
 254      y=c(15),
 255      labels=c('Release', '3.0', '3.1', '', '3.2', '3.3', '4.0', '4.1', '4.2', '4.3'))
 256
 257 #tstamp_20041112 <-abline_at_date("2004-11-12", col='grey60', lty=2, height=10)
 258 #tstamp_20050301 <-abline_at_date("2005-03-01", col='grey60', lty=2, height=10)
 259 #tstamp_20050615 <-abline_at_date("2005-06-15", col='grey60', lty=2, height=10)
 260 #tstamp_20051023 <-abline_at_date("2005-10-23", col='grey60', lty=2, height=10)
 261 #tstamp_20070101 <-abline_at_date("2007-01-01", col='grey60', lty=2, height=10)
 262 #tstamp_20070501 <-abline_at_date("2007-05-01", col='grey60', lty=2, height=10)
 263 #tstamp_20080601 <-abline_at_date("2008-06-01", col='grey60', lty=2, height=10)
 264 #tstamp_20080815 <-abline_at_date("2008-08-15", col='grey60', lty=2, height=10)
 265 #tstamp_20090501 <-abline_at_date("2009-05-01", col='grey60', lty=2, height=10)
 266 #tstamp_20100201 <-abline_at_date("2010-02-01", col='white', lty=2, height=10)
 267 #
 268 #
 269 #text(x=c( tstamp_20041112+(tstamp_20050301-tstamp_20041112)/2,
 270 #        tstamp_20050301+(tstamp_20050615-tstamp_20050301)/2,
 271 #        tstamp_20050615+(tstamp_20051023-tstamp_20050615)/2,
 272 #        tstamp_20051023+(tstamp_20070101-tstamp_20051023)/2,
 273 #        tstamp_20070101+(tstamp_20070501-tstamp_20070101)/2,
 274 #        tstamp_20080601+(tstamp_20080815-tstamp_20080601)/2,
 275 #        tstamp_20090501+(tstamp_20100201-tstamp_20090501)/2 ),
 276 #     y=c(15),
 277 #     labels=c('3.0', '3.1', '3.1S', '3.2', '4.0', '4.2', '4.3'))
 278
 279 end_image()