Add initial support for chrony statistics.
At this point introduce time offset and stratum stats.
There is something strange with the host option.
Anything else than 'localhost' or '127.0.0.1' causes an error at
collectd start, even when /tmp/collectd/conf is directly edited
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
Shorten the visible delay at the statistics graph endpoint
from 60 seconds to 15 seconds.
The 60s delay was recently added as a way to remove a possible
visible gap at the end of the hourly graph due to absence of
recent data. Default data collection interval is 30s, so that
60s guarantees that there is data upto to the end of the graph.
However, that 60s makes any recent activity to get displayed
really slowly, after a 60-89 second delay.
Shortening the gap to 15s, half of the default data collection
period, should balance things:
* Half of the time there may be a really narrow gap visible and
half of the time there is no gap at all.
* The most recent displayed data point is from 15-44 seconds ago
(instead of 60-89 seconds ago).
Neither 15 or 60 seconds makes any impact to the longer graphs.
To accommodate to possibly shorter timespans, and to avoid the
occasional wrong data series selected for longer period graphs
(see #4065) the endpoint delay might be relative to the timespan.
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
Add clarifying help text to the RRD time span definition.
Only a few strings are allowed and it is easy to cause uninteded
results especially with minutes as only "min" is for minute while
("m" or "month") are for month.
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
Adjust the defaults shown in the LuCI user interface to match
the real default values in the default config file.
(If a plugin is disabled and config values get deleted from
the config file, user has been offered incorrect default
values from UI defaults when the plugin is later re-enabled.)
* dns: set br-lan as the interface
* email: socket in /var/run/collectd/ dir
* interface: set br-lan as the interface
* ping: TTL 127, interval 30
* rrdtool: 1hour as shortest period, 144 RRArows
* tcpconns: Do not monitor all, only 22 80
* unixsock: socket in /var/run/collectd/ dir
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
Add the dhcpleases plugin:
https://github.com/openwrt/packages/pull/14204
Added-Frontend-by: Hannu Nyman <hannu.nyman@iki.fi>
Signed-off-by: Nick Hainke <vincent@systemli.org>
[Minor changes to graph: alt_autoscale, precision, label]
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
* Add functionality to hide the metric of 'idle' state
from the chart and data table. Many routers are mostly idle,
and the 98% 'idle' dominates the graph reducing its usefulness.
Without 'idle', the smaller CPU usage spikes are visible.
* Hide 'idle' by default. Provide config option to show it.
(note: the option in inside LuCI, and has no impact on actual
data collection by collectd.)
* Tweak the defaults to use the percentage data by default.
It makes more sense to average users than jiffies.
* Set the current LuCI defaults also in the config file.
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
It is sometimes useful not to display avg values int the graph.
This option is used to prevent this from appearing in the graph.
In a graph definition this option must be set to *true*,
if the avg values should not be displayed.
```
options = {
gauge = {
title = "Status",
color = "0000ff",
noarea = true,
noavg = true
}
```
Signed-off-by: Florian Eckert <fe@dev.tdt.de>
Apparently " in a comment may cause an error in some conditions.
Reference to issue #4020
Remove also ";" just in case...
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
Add graphs for sqm data collected by 'sqm_collect.sh'
An overall sqm graph displays bandwidth (Kb/s), Backlog (Bytes) and
drops for each supported qdisc.
Sqm-cake supports cake's tin structure and displays many(!) graphs per
tin:
Bandwidth: Bandwidth (Kb/s), bandwidth threshold (Kb/s).
Latency: Target latency, Peak, Average, Sparse flow latency
Backlog v Drops: Backlog (Bytes), Drops, ECN marks, Ack-filter drops
Flow counts: Sparse flows, Bulk flows, Unresponsive flows
Fun for all the family :-)
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
The existing graph shows a conntrack entry usage value but doesn't show
a conntrack table size figure so you don't know if you're close to
filling the table. Add a percent usage graph to show conntrack table
percent used.
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
- Add a per-plugin overview tab that contains the (non-detail) graphs of
each plugin instance, similar to the old Lua based implementation
- Numerically order plugin instances starting with numbers
- Avoid multiple renderings of the same graphs
- Fix legend of tcpconns graphs
- Move cpufreq distribution and transition charts to detail tabs
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
- Remove redundant form flags
- Fix various default values
- Add some new collect options and remove options that do not exist anymore
Signed-off-by: Jo-Philipp Wich <jo@mein.io>
