Ah I C ! Are the log files under each stored builds large? If so, this could mean that the builds are updating the logfiles with too much data.
Are there any dirs (containing logfiles) under the builds directory that should have been auto-deleted, based on the job configuration setting to only keep a certain number of builds? I hope that my questions may lead to resolving your problem. Kind Regards, Marek On Fri, Sep 21, 2012 at 10:49 AM, Miguel Almeida <[email protected]>wrote: > Hi Marek, > > On Fri, Sep 21, 2012 at 3:36 PM, Marek Gimza <[email protected]> wrote: > >> Miguel, >> > > Is the workspace directory under the >> /usr/share/tomcat6/.jenkins/jobs/<job-name> directories? >> > It is. But the workspace directly below <job-name> is only 100 MB large, > so it's hardly the problem. > Running some "du" commands, I see most space is being occupied by the > builds directories under each module: > /usr/share/tomcat6/.jenkins/jobs/<job-name>/moduleA/builds. One of them > has...15GB of data! > > >> This could be the reason for the disk usage. >> The workspace is the directory to which jenkins will sync and perform >> your build-steps. >> >> >> You could take advantage of the "customWorkspace" field in the job >> configuration or the "Remote FS root" field in the Node's configuration to >> specify a different directory to store the workspace. >> >> We use these fields to specify our workspace for each job to be different >> than the job's meta-data, such as log-files, which is still under the >> ${JENKINS_HOME}/jobs/<job-name> directories. >> >> >> Kind Regards, >> Marek >> >> >> >> On Fri, Sep 21, 2012 at 10:26 AM, Miguel Almeida < >> [email protected]> wrote: >> >>> Dear all, >>> >>> I have been using Jenkins for some months now and I am interested in the >>> issue of disk usage. >>> >>> While trying to understand why the 50GB on the server were becoming >>> short, I decided to investigate the size of each job directory under >>> /usr/share/tomcat6/.jenkins/jobs/. To my surprise, this was larger than >>> expected. One maven job with 5 modules and about 700 runs is currently >>> taking 16 GB of disk space! >>> >>> I realize I can "discard old builds" of a job, but then I'll lose >>> interesting metrics like code coverage trends or test result trends. My >>> questions are: >>> >>> 1) Is this a normal usage - 23-ish MB per job run? >>> 2) If so, are there other options that allow me to keep a relatively >>> interesting history but without taking so much disk space? >>> 3) Is this a usual concern, or do you just splash new TB disks whenever >>> you run out of space? I mean, I've been using Jenkins for 10 months and >>> have around 20 projects now, surely this is not intensive usage. >>> >>> >>> I appreciate the feedback, >>> >>> Miguel Almeida >>> >> >> >
