Red Hat Bugzilla – Bug 1293360
When filesystem name is long, the disk full check fails to work
Last modified: 2016-08-30 08:14:05 EDT
+++ This bug was initially created as a clone of Bug #1196774 +++ Description of problem: The logwatch has a perl file zz-disk_space which contains a function DiskFull. This function does a split to separate lines, mostly using the df command. When the filesystem name is long, it appears on a line by itself, with the other fields (Size, Used, Avail, Use%, Mounted on) appearing on the next line. Because of this, the check for the filesystem becoming full looks at the wrong field, and fails to report when the disk is becoming full. Version-Release number of selected component (if applicable): logwatch-7.4.0-28.20130522svn140.el7.noarch How reproducible: Very Steps to Reproduce: 1. Go to a host which has a long (> 22 chars) filesystem name. 2. Make its disk usage at or above 90%. Create a big file with truncate or fallocate. 3. sudo logwatch --detail high Actual results: No 'Warning. Disk filling up.' message appears. Expected results: Something like this: /dev/mapper/vg_gw-lv_root => 100% Used. Warning. Disk Filling up. Additional info:
I took a closer look into the matter and was unable to reproduce this bug with the latest released version (logwatch-7.4.0-28.20130522svn140.el7.noarch). It turns out that from coreutils 8.11 on, lines with long filesystem names are no longer being split in two: df now aligns columns consistently, and no longer wraps entries with longer device identifiers, over two lines. (From http://savannah.gnu.org/forum/forum.php?forum_id=6785) This means that disk usage info is located consistently (5th column) and logwatch always prints the 'Disk Filling up' warning when needed, even for filesystems with monstrously long names. As there is nothing to fix here, I am closing the bug.
I'm not sure why this bug report got resubmitted. The problem was fixed by Jan Synacek on 2015-11-19 02:56:42 EST under bug 1196774.