cpumemlogplot

#!/usr/bin/env Rscript

### cpumemlogplot
###----------------------------------------------------------------------------

args <- commandArgs(TRUE)

help <- function()
{
  cat("

  Summarize and plot cpumemlog.sh output

  Usage:
    cpumemlogplot.R [argument(s)] [file(s)]

  Arguments:
    --help      - print this text

    --mega,-m   - plot memory usage in MB (GB is default)

    --group,-g  - group process's data by process (command) name

    --labels,-l - use directlabels to position process labels onto lines
                  instead of using a legend

  Example:
    Summarize and plot output stored in file ...

      cpumemlogplot.R cpumemlog_5026.txt

    Summarize and plot output - show RAM usage in megabytes

      cpumemlogplot.R --mega cpumemlog_5026.txt

    Summarize and plot output stored in files ...

      cpumemlogplot.R cpumemlog_5026.txt cpumemlog_1000.txt

  Author
    Gregor Gorjanc <gregor dot gorjanc at gmail dot com>

  Web
    http://github.com/gregorgorjanc/cpumemlog\n\n")
  q(save="no")
}

if(length(args) < 1 | "--help" %in% args) {
  help()
}

binPower <- 20
binUnit <- "Gb"
test <- c("--mega", "-m") %in% args
if(any(test)) {
  args <- args[!(args %in% c("--mega", "-m"))]
  binPower <- 10
  binUnit <- "Mb"
}

test <- c("--group", "-g") %in% args
if(any(test)) {
  args <- args[!(args %in% c("--group", "-g"))]
  group <- TRUE
} else {
  group <- FALSE
}

test <- c("--labels", "-l") %in% args
if(any(test)) {
  args <- args[!(args %in% c("--labels", "-l"))]
  labels <- TRUE
} else {
  labels <- FALSE
}

# Colection of colors from ColorBrewer
.color <- c("#E41A1C", "#377EB8", "#4DAF4A", "#984EA3", "#FF7F00", "#FFFF33",
            "#A65628", "#F781BF", "#999999", "#1B9E77", "#D95F02", "#7570B3",
            "#E7298A", "#66A61E", "#E6AB02", "#A6761D", "#666666", "#7FC97F",
            "#BEAED4", "#FDC086", "#FFFF99", "#386CB0", "#F0027F", "#BF5B17",
            "#A6CEE3", "#1F78B4", "#B2DF8A", "#33A02C", "#FB9A99", "#E31A1C",
            "#FDBF6F", "#CAB2D6", "#6A3D9A", "#B15928", "#66C2A5", "#FC8D62",
            "#8DA0CB", "#E78AC3", "#A6D854", "#FFD92F", "#E5C494", "#B3B3B3",
            "#8DD3C7", "#FFFFB3", "#BEBADA", "#FB8072", "#80B1D3", "#FDB462",
            "#B3DE69", "#FCCDE5", "#D9D9D9", "#BC80BD", "#CCEBC5", "#FFED6F")

options(width=300)

library(package="doBy")
library(package="reshape2")
library(package="ggplot2")

if(labels) {
  library(package="directlabels")
}

for(file in args) {
  ## file <- args[1]

  cat("File:", file, "\n")

  ## --- Handle data ---

  dat <- read.table(file=file, header=TRUE, fill=TRUE)
  cols <- c("PID", "PCPU", "PMEM", "VSZ", "RSS")
  dat[, cols] <- lapply(dat[, cols], as.numeric)
  sel <- grepl(pattern="defunct", x=dat$DATE)
  dat <- dat[!sel, ]
  dat$X <- as.POSIXct(paste(dat$DATE, dat$TIME))
  dat$RSS <- dat$RSS / 2^binPower
  if (!group) {
    dat$COMMAND <- paste(dat$COMMAND, as.numeric(factor(paste(dat$COMMAND, dat$PID, sep="_"))), sep="_")
  }


  ## Summarize
  sink(file=paste(file, "mem_stat.txt", sep="_"), split=TRUE)
  cat(paste("\nRAM - RSS (", binUnit, "):\n", sep=""))
  tmp <- summaryBy(RSS ~ 1, data=dat, FUN=descStat)
  print(tmp)
  cat(paste("\nRAM - RSS by command (", binUnit, "):\n", sep=""))
  tmp <- summaryBy(RSS ~ COMMAND, data=dat, FUN=descStat)
  print(tmp)
  sink()

  sink(file=paste(file, "cpu_stat.txt", sep="_"), split=TRUE)
  cat("\nCPU:\n")
  tmp <- summaryBy(PCPU ~ 1, data=dat, FUN=descStat)
  print(tmp)
  cat("\nCPU by command:\n")
  tmp <- summaryBy(PCPU ~ COMMAND, data=dat, FUN=descStat)
  print(tmp)
  sink()

  sink(file=paste(file, "time_stat.txt", sep="_"), split=TRUE)
  cat("\nTIME:\n")
  dat$TIME <- dat$X
  tmp <- summaryBy(TIME ~ 1, data=dat, FUN=function(x) c(min=min(x), max=max(x)))
  ## TODO: is this origin stuff portable? Could we get origin from dat$X somehow?
  tmp[, ] <- lapply(tmp[, ], FUN=as.POSIXct, origin="1970-01-01")
  tmp$TIME.diff <- tmp$TIME.max - tmp$TIME.min
  print(tmp)
  cat("\nTIME by command:\n")
  tmp <- summaryBy(TIME ~ COMMAND, data=dat, FUN=function(x) c(min=min(x), max=max(x)))
  tmp[, -1] <- lapply(tmp[, -1], FUN=as.POSIXct, origin="1970-01-01")
  tmp$TIME.diff <- tmp$TIME.max - tmp$TIME.min
  print(tmp)
  sink()

  ## --- Plot ---

  p <- ggplot(aes(x=X, y=RSS, colour=COMMAND), data=dat)
  p <- p + geom_line()
  p <- p + xlab(label="Time")
  p <- p + ylab(label=paste0("RAM - RSS (", binUnit, ")"))
  p <- p + scale_color_manual(values=.color)

  png(file=paste(file, "mem_plot.png", sep="_"), width=480*1.5, height=480)
  if(labels) {
    print(direct.label(p, method="last.points"))
  } else {
    print(p)
  }
  dev.off()

  p <- ggplot(aes(x=X, y=PCPU, colour=COMMAND), data=dat)
  p <- p + geom_line()
  p <- p + xlab(label="Time")
  p <- p + ylab(label="CPU (%)")
  p <- p + scale_color_manual(values=.color)
  png(file=paste(file, "cpu_plot.png", sep="_"), width=480*1.5, height=480)
  if(labels) {
    print(direct.label(p, method="last.points"))
  } else {
    print(p)
  }
  dev.off()

}

###----------------------------------------------------------------------------
### cpumemlogplot ends here