I have hesitated to report this because, as Texstar has pointed out, e17 is under heavy development
. In any case, I've been very happy with Texstar's packages. This e17 install is, by far, the most stable I have ever used on any distro. Anyway, here goes:
I ran htop the other day on my e17 desktop to check that my folding clients were at the top of their game. They were both running in the 90% range of CPU utilization. It was then that I noticed a single task was using over 60% of my 4GB of RAM. The task was this one:root 11100 1.1 33.3 1379444 1368664 tty9 Ss+ Jul13 52:59 /etc/X11/X :0 -audit 0 -deferglyphs 16 -auth /var/lib/gdm/:0.Xauth -nolisten tcp vt9
Running ps aux | grep gdm
shows:root 1004 0.0 0.0 17868 3392 ? S Jul09 0:00 /usr/sbin/gdm-binary -nodaemon
root 1035 0.0 0.0 18860 3628 ? S Jul09 0:00 /usr/sbin/gdm-binary -nodaemon
root 11100 1.1 33.3 1379444 1368664 tty9 Ss+ Jul13 52:59 /etc/X11/X :0 -audit 0 -deferglyphs 16 -auth /var/lib/gdm/:0.Xauth -nolisten tcp vt9
darrel 20641 0.0 0.0 3268 752 pts/2 SN+ 00:28 0:00 grep --color gdm
So, there are two separate gdm-binary tasks running. I then checked this on two PCLOS 2010 LXDE installs, and one Zen-mini 2010 install. The same four tasks are running in each case. However, none of those other desktops have the memory leak problem. This seems to be an e17-specific issue. Why, I don't know. The only way to reclaim the lost memory seems to be to restart X. The three-finger salute, Ctrl-Shift-End, to restart e17, has no effect on the memory loss. It simply restarts the e17 desktop in about two seconds. The screenshots below were taken approximately 24 hours apart over a five day period, and show the cumulative effect of the memory loss.
The free -t
command, taken before and after restarting X, shows the memory used and reclaimed.
Before logging out: total used free shared buffers cached
Mem: 4101156 3787588 313568 0 584 1659904
-/+ buffers/cache: 2127100 1974056
Swap: 2096440 5660 2090780
Total: 6197596 3793248 2404348
After logging in: total used free shared buffers cached
Mem: 4101156 2037096 2064060 0 584 1703576
-/+ buffers/cache: 332936 3768220
Swap: 2096440 5164 2091276
Total: 6197596 2042260 4155336
Am I the only one seeing this behavior? OK, I'm ducking and running for cover now.