I've collected some data about disk usage while doing last testing build. Here you can see it: http://bit.ly/dnzIrq Or long direct url: https://spreadsheets.google.com/ccc?key=0AhY8lmfkCabTdGZRSGZ0eXg5cWhkV2k1c0Jvd0lvVnc&authkey=CLeK150N#gid=0
Google Spreadsheet chart loading will take some time so be patient :) Columns A,B contains full measured data. Columns C,D are selection from A,B (every 10th record) to reduce chart drawing time. Resulting view is almost the same as with full data. Disk usage values are 'real' Mb (/1024). Data was collected from following build: BB version: master OE version: testing_2010-10-29 rm_work: enabled Parallelism: 3 BB threads + 3 make threads Distro: angstrom-2008.1 Machines: efikamx (first peak), ben-nanonote (second peak) Images: console-image x11-image opie-image Build type: clean, but with different TMPDIR's for every machine w/o removing of TMPDIR after build. All sources was downloaded before and will not affect disk usage data. Disk usage was measured by storing value of 'Available' column of 'df -Pk $TMPDIR' every 10 seconds. Both TMPDIRs are on same partition. There is no other processes who can write to this partition. After collecting data I've found minimal disk usage value and substract if from all values. Note that ben-nanonote values include size of efikamx TMPDIR after build. You can see here following things: 1) peak disk usage is about 23Gb 2) rm_work is working (right after peak), but, imho, is starting a bit late :) Any comments/suggestions/requests here? :) -- Yury Bushmelev _______________________________________________ Openembedded-devel mailing list [email protected] http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
