Sam Varshavchik wrote:
Jim Cornette writes:
>
> The problem is encountered on a laptop which is shutdown frequently
> after use.
> When the library.so.n.#prelink#<random-characters> file is
> encountered, there is usually the library.so.n file intact with an
> earlier date for it. When the #prelink# files are located, usually
> there are entries for the same library with dates either a day or two
> later for the #prelink# version. This problem is bothersome. However,
> the problem with library symlinked to /tmp/prelink<random-characters>
> scares me since I clean /tmp frequently and realize that others use
> ramdisks for files destined to /tmp. Prelink might be better off
> using var/cache/prelink for its temporary action for in transition
> actions.
During shutdown all existing processes should receive a SIGTERM first,
then SIGKILL five seconds later if they are still running.
File a bug against prelink, stating that prelink should be catching the
SIGTERM, and cleaning up after itself.
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=176570
Would the 5 seconds before a SIGKILL be a smart idea in modern Linux?
This practice seems barbaric! If a process is not showing any activity,
it might need killed, but killed in process?
Thanks!
Jim
--
"Nuclear war can ruin your whole compile."
-- Karl Lehenbauer