logrotate(8) and copytruncate as default

Lennart Poettering mzerqung at 0pointer.de
Fri Jun 28 13:23:11 UTC 2013

On Fri, 28.06.13 13:23, P J P (pj.pandit at yahoo.co.in) wrote:

> > The systemd-journald takes care of all of: receiving messages, writing
> > them to storage, and rotating the storage.
> >
> > We do synchronous rotation before each write. i.e. the moment we append
> > to a file we check if the write would cause the disk usage to be out of
> > limits, and then do the rotation right away.
>    I see.  While doing this rotation, I guess systemd uses flock(2) or similar
> mechanism to pause writing to a log file, move/rename or copy-truncate that
> file and continue writes again?

journald is the only writer, it doesn't need locking. The changes it
does are done in a way so that concurrent readers will either see the
changes or not, but never half-written changes. 

Also note that locking on Linux is seriously broken. You can get a lock
on any file you can read, thus you can freeze everybody else who might
want a lock. Or in other words: if a logging daemon takes a lock on its
lock files, then you can use this to make that service freeze forever.

File system locking on Unix cannot really work. 


Lennart Poettering - Red Hat, Inc.

More information about the devel mailing list