The system statistics collection daemon. Mirror of the canonical repository (updated hourly), but feel free to send Pull Requests!
Fetching latest commit…
Cannot retrieve the latest commit at this time
collectd - System information collection daemon ================================================= http://verplant.org/collectd/ About ----- collectd is a small daemon which collects statistics about a computer's usage and writes then into RRD files. Features -------- * collectd is able to collect the following data: - CPU utilization (Time spent in system, user, nice and idle) - CPU frequency (For laptops with speed step or a similar technology) - Mountpoint usage (Basically the values `df(1)' delivers) - Disk utilization (Sectors read/written, number of read/write actions, time spent doing IO) - Harddisk temperatures (Uhm, yeah, temperature of harddisks that is ;) - System load (Load average over the last 1, 5 and 15 minutes) - Memory utilization (Memory occupied by running processes, page cache, buffer cache and free) - MySQL server statistics (Commands issued, handlers triggered, thread usage, query cache utilization and traffic sent/received) - NFS Procedures (Which NFS command were called how often. Only NFSv2 and NFSv3 right now) - Ping latency (Time to reach the default gateway or another given host) - Process counts (Number of running, sleeping, zombie, ... processes) - Serial (RX and TX of serial interfaces) - Sensors (System temperatured and fan rotation speeds) - Swap (Pages swapped out onto harddisk or whatever is called `swap' by the OS..) - Tape (Read and write bytes and operations on tape devices) - Traffic (In/Outbound traffic on the interfaces) - Users (Currently logged in users) * Performance: Running as a daemon collectd doesn't spend much time in startup. Since collectd links against libping, librrd and libsensors it doesn't need to start any other processes. * Hardly any maintenance neccessary and setup is trivial. * Extremely easy and failsafe network operation possible. Operation --------- * collectd's configuration file can be found at `sysconfdir'/collectd.conf. Run `collectd -h' for a list of builtin defaults. See `collectd.conf(5)' for a list of options and a syntax description. * When running collectd writes system statistics in RRD-files. Per default they reside in `/var/lib/collectd'. * When using the `ping' plugin collectd needs to run as user root, since only root can craft ICMP packages needed to ping other hosts. collectd should NOT be installed setuid root since it can be used to overwrite valuable files.. * Sample scripts to generate graphs reside in `contrib/' in the source package or somewhere near `/usr/share/doc/collectd' in most distributions. Please be aware that those script are meant as a starting point for your own experiments.. Some of them require the `RRDs' Perl module. (`librrds-perl' on Debian) * The RRD-files that collectd creates hold the following data. Use ``rrdtool resize'' if you want to cover longer/shorter periods of time. Resolution | Data points | Timespan -----------+-------------+---------- 60 seconds | 1500 | 25 hours 30 minutes | 1680 | 35 days 6 hours | 1520 | 380 days Prerequisites ------------- To compile collectd from source you will need: * Usual suspects: C compiler, linker, preprocessor, make, ... * rrdtool (headers and library; rrdtool 1.0 and 1.2 both work fine) If built without `librrd' the resulting binary will be `client only', i.e. will send it's values via multicast and not create any RRD files itself. * libmysqlclient (optional) * lm-sensors (optional) * libstatgrab may be used to collect statistics on systems other than Linux and/or Solaris. Note that CPU- and disk-statistics, while being provided by this library, are not supported in collectd right now.. <http://www.i-scream.org/libstatgrab/> Author ------ Florian octo Forster <octo at verplant.org>