Martin Meredith wrote:
> I've had a problem where somehow, I've managed to end up with
> approximately 1,000,000 session files on my server.
That is definitely a very large number of files all in one place!
> Due to the large amount of files, the current crontab to clear them
> was unable to deal with it (xargs would fail to take in the HUGE
> list of files).
>
> It seems that rather than using xargs (even with the limit), that
> using the -exec option of find might be a little bit more sane?
I realize that the cron script has already changed in later releases
but just the same this shouldn't have been a problem with the previous
version that used xargs. Using xargs is an older but quite acceptable
way to deal with a very large list of filenames. I think that there
was some different problem related to this that wasn't diagnosed.
In other words, while using 'find . -exec command {} +' is more
efficient than 'find . -print0 | xargs -0 command' they are both
equivalent in functionality. And using 'find . -conditions -delete'
is even better for safety and efficiency the 'xargs -0' version is
still functionally correct. Today it is best to use -delete since it
is newly available in find. But before the -delete and the "{} +"
forms were available then the 'find . -print0 | xargs -0 ...' form was
the standard of excellence and will handle directories with millions
of files in it.
In yet other words, I feel certain that the actual problem was
something else and not the use of 'find . ... -print0 | xargs -0 ...'
there. For example perhaps memory was so limited that the Linux
kernel out-of-memory killer (OOM Killer) came into place? Just as an
example and not necessarily saying that was the specific problem you
hit.
Bob
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
