Now with the patches integrated to filter logging output, we can finally
remove forking a subprocess and stop redirecting file descriptors.
We instead use exec to replace the process with the final daemon.
Signed-off-by: Kai Krakow <kai@kaishome.de>
To make bees more friendly to use with syslog/systemd, we add an option
to omit timestamps from the log output.
Signed-off-by: Kai Krakow <kai@kaishome.de>
This commit adds a simple getopt options parser to show help. This can
be used as a boilerplate for adding more options later.
Signed-off-by: Kai Krakow <kai@kaishome.de>
To install for different distributions, LIBEXEC_PREFIX can now be set.
It defaults to $(PREFIX)/usr/lib/bees as used in most common
distributions.
Local overrides are possible by setting variables in a "localconf" file
which will be included by the Makefile if it exists.
For some distributions you may want to set it to /usr/libexec or
/usr/libexec/bees.
Let's remove the CPUQuota example and instead give bees a share of
what's available.
128 CPU shares will give it about 12% max CPU under load, give it a
slight boost during startup to allow reading the hash table faster.
100 block shares will give it about 10% max disk bandwidht under load,
give it a slight boost during startup to allow reading the hash table
faster.
Then let's adjust the CPU and IO scheduler to prefer other processes.
This way bees runs completely in the background, barely noticable
during, e.g., gaming.
Explicitly set control-group kill mode, that is: try SIGTERM first, and
use SIGKILL after a timeout. This exactly defines how bees is running as
a child process within the frontend service starter. Not sure if bees cares
about signals but SIGTERM first seems cleaner. On the way, let bees restart
on abnormal termination.
If you have a lot of or a few big nocow files (like vm images) which
contain a lot of potential deduplication candidates, bees becomes
incredibly slow running through a lot "invalid operation" exceptions.
Let's just skip over such files to get more bang for the buck. I did no
regression testing as this patch seems trivial (and I cannot imagine any
pitfalls either). The process progresses much faster for me now.
Kai Krakow