Revision as of 22:16, 17 October 2011

Analizing OpenStack code

Tools

• cvsanaly
• MySQL
• Pentaho Report
• GNU R

Installing cvsanaly2 from https://projects.libresoft.es/projects/cvsanaly/wiki Clone all the repositories listed on https://github.com/openstack/:

$ git clone git://.... [glance|swift|nova...]

Then create the databases for CVSanaly2 to store the logs in

$ mysqladmin -u mlstats -p create openstackrepos

And run cvsanaly2 on each of the repositories:

~/projects/swift$ cvsanaly2 --extensions Metrics,FileTypes,CommitsLOC -u mlstats -p xxxxxxx -d openstackrepos
Parsing log for swift/
Executing extensions
Executing extension FileTypes
Executing extension Metrics

Using GNU R to produce graphs

After starting R, connect to MySQL database:

library(RMySQL)
con <- dbConnect(dbDriver("MySQL"), user="mlstats", password="xxxxx", dbname="openstackrepos")

Number of commits per month

query <- "SELECT date_format(s.date, '%m/%Y') date, count(s.id) commits
          FROM scmlog s group by date_format(s.date,'%Y%m');"
results <- dbGetQuery(con,query)
evol_commits <- ts(results$commits, start=c(2010,7), freq=12)
  png("nova_commits_month.png")
plot(evol_commits, type="l", xlab="Date", ylab="Commits",
     main="Number of commits per month - NOVA")
query_avg <- "SELECT AVG(g.numcommits)
    FROM
    ( SELECT date_format(s.date, '%Y') myyear,
        date_format(s.date, '%m') mymonth, count(s.id) numcommits
        FROM scmlog s
        GROUP BY date_format(s.date,'%Y%m') ) g;"
result_avg <- dbGetQuery(con,query_avg)
qqline(result_avg, col="blue", lty=2)
query_max_min <- "SELECT MAX(g.numcommits) as max, MIN(g.numcommits) as min
    FROM
    ( SELECT date_format(s.date, '%Y') myyear,
        date_format(s.date, '%m') mymonth,
        count(s.id) numcommits
        FROM scmlog s
        GROUP BY date_format(s.date,'%Y%m') ) g;"
result_max_min <- dbGetQuery(con,query_max_min)
qqline(result_max_min$max, col="red", lty=2)
qqline(result_max_min$min, col="green", lty=2)
legend("topright", inset=.05, c("average","maximun","minimun"),
       fill=c("blue","red","green"))
dev.off()

Number of Commits per author

query <- "SELECT p.name author, count(s.id) commits
    FROM scmlog s LEFT JOIN people p ON s.committer_id=p.id
    GROUP BY committer_id ORDER BY commits;"
results <- dbGetQuery(con,query)
plot(results$commits, xlab="Author", ylab="Commits",
     main="Number of commits by author")

Number of commits by author per year

query <- "
SELECT year, name, num
FROM
(SELECT date_format(s.date, '%Y') AS year, p.name AS name, count(s.id) AS num
FROM scmlog s LEFT JOIN people p ON s.committer_id=p.id
GROUP BY year, name
ORDER BY year, num) g
WHERE g.num > 10;
"
results <- dbGetQuery(con,query)
results$year <- factor(results$year)
query_names <- "
SELECT DISTINCT(name)
FROM
(SELECT date_format(s.date, '%Y') AS year, p.name AS name, count(s.id) AS num
FROM scmlog s LEFT JOIN people p ON s.committer_id=p.id
GROUP BY year, name
ORDER BY year, num) g
WHERE g.num > 10;
"
names <- dbGetQuery(con,query_names)
col <- 1
for (i in names$name) {
    results$color[results$name==i] <- col
    col <- col + 1
}
dotchart(results$num, groups=results$year, labels=results$name,
         color=results$color, cex=.7, xlab="Number of commits",
         main="Commits by author per year")

Aggregated number of commits up to now

query <- "SELECT g.myyear, g.mymonth, g.numcommits,
        (@sumacu:=@sumacu+g.numcommits) aggregated_numcommits
    FROM
        (SELECT @sumacu:=0) r, (SELECT date_format(s.date, '%Y') myyear,
            date_format(s.date, '%m') mymonth,
            COUNT(s.id) numcommits
        FROM scmlog s
        GROUP BY date_format(s.date,'%Y%m')) g;"
results <- dbGetQuery(con,query)
evol_num_commits <- ts(results$aggregated_numcommits, start=c(2010,10), freq=12)
plot(evol_num_commits, type="h", xlab="Date", ylab="Commits",
     main="Aggregated number of commits", col = "dark blue")

In order to save an image of the plot use

  png("<filename>.png")
  plot <command>
  dev.off()