Paul Cuzner [Fri, 7 Jul 2017 04:01:50 +0000 (16:01 +1200)]
dashUpdater: remove $domain from dashboards, if domain is not configured
For environments that don't use dns, collectd will not provide a FQDN
on the metric name. In these circumstances, the dashboards are empty.
This fix looks for the domain setting, and if it's not supplied the
$domain reference in all queries is removed before the dashboard is loaded
into grafana.
Paul Cuzner [Thu, 6 Jul 2017 23:31:48 +0000 (11:31 +1200)]
osd: add support for osd related stats, and support journal devices
OSD daemons are now asked for perf data, so latencies within ceph can be
loaded to graphite. In addition the journal device is detected. If it's
not collocated on the osd device, additional disk metrics under a journal
subtree are created within graphite
Paul Cuzner [Thu, 6 Jul 2017 23:29:06 +0000 (11:29 +1200)]
common: changes to the Disk class
Two main things;
1. Disk instances are now initialized here, instead of with the caller
devices simplying code in the osd class
2. get_real_dev function added to convert a device name of an OSD to the
name we'll use as a metric. this now provides initial support for nvme
and intelcas based osd
Paul Cuzner [Fri, 30 Jun 2017 02:05:33 +0000 (14:05 +1200)]
at-a-glance: multiple fixes to mon/osd/growth and forecast panels
MON/OSD panel queries updated to address the interpolation
problem where floats were shown. OSD panel also now shows
total OSDs
Templating update for the disk_full_threshold (2->80)
Growth/Forecast panel queries updated to account for data coming
from multiple mon's
Health Panel updated to show as RED when the cluster is in an
ERROR state
Paul Cuzner [Thu, 29 Jun 2017 04:50:30 +0000 (16:50 +1200)]
at-a-glance: pg status pie chart changes
a degraded state is now shown based on the diff of pg_active and
pg_active_clean. This intermediate metric has been added to the pie
chart so it shows; active+clean, degraded and peering.
Paul Cuzner [Thu, 29 Jun 2017 03:15:58 +0000 (15:15 +1200)]
network-usage: dashboard updated to track enX interface stats
graphite doesn't support blacklisting in queries, so interface names that
we're interested in have to be whitelisted. This fix now tracks enX, ethX and
bondX interface names.
Paul Cuzner [Thu, 29 Jun 2017 03:12:54 +0000 (15:12 +1200)]
ceph-rados: display fixes to a several charts
health history was a mess with the light theme. The chart now uses threshold lines
(amber and red), and plots health against those lines. In addition small fixes to the capacity
chart (it was stacking values!), and the monitor status table
Paul Cuzner [Thu, 29 Jun 2017 03:09:01 +0000 (15:09 +1200)]
backend-storage: cosmetic changes to heatmap and graphs
Heatmap 'spectrum' was not showing well in the light theme - this
make it more readable. In addition 'info' has been added to explain the
heatmap and what it represents.
Paul Cuzner [Thu, 29 Jun 2017 03:04:17 +0000 (15:04 +1200)]
at-a-glance : query and cosmetic changes
Multiple changes as follows;
- dashboard links (top) hover over was mis-aligned. fixed
- forecast value if negative now shows N/A
- forecast and growth queries updated
- disks near full set to 0, if there aren't any issues (instead of no value)
- added descriptions on various panels
Paul Cuzner [Thu, 29 Jun 2017 02:53:43 +0000 (14:53 +1200)]
status-panel: Update the bkgnd color to support the light theme
By default the panel just uses 'green', which with the light theme is too
dark, making the text on the panel difficult to read. This ansible step just
updates the color in the css to make the text more readable
Boris Ranto [Mon, 26 Jun 2017 12:30:38 +0000 (14:30 +0200)]
collectors: Pass through keyword arguments
We need this to pass through the log_level keyword argument to the base
class. Otherwise, the collectd will fail because these classes get
unknown argument log_level.
Boris Ranto [Sun, 25 Jun 2017 08:13:51 +0000 (10:13 +0200)]
ansible: Write grafana config when grafana is down
We need to start and communcite with grafana after we push our own
config to the grafana. The old config can use different locations
e.g. for the DB and these won't get populated properly if we run
dashUpdater or post to the grafana API too early.
Paul Cuzner [Mon, 26 Jun 2017 05:33:45 +0000 (17:33 +1200)]
at-a-glance: various enhancements
Dashboard updates;
- initial light theme support - some colour changes on charts
- shuffled panel order so the rows basically cover - overview, client and OS
- added disks near full panel
- added growth and forecast panels
- added OSD level ram usage summary
- PG's now shown using Grafana labs pie chart plugin
- count of rbds now shown (on client row)
- osd host count query changed
- queries on Mons and OSDs changed to mitigate state flapping
- 'buttons' changed to make them work with light/dark themes
Paul Cuzner [Mon, 26 Jun 2017 05:12:23 +0000 (17:12 +1200)]
mon: updated for logging and additional metrics collected for rbd and osd hosts
collector class now scans the cluster to determine the count of RBDs. Each monitor
will pick a discrete set of pools to scan, so the overall load is shared across monitors.
In addition, since the osd tree command is used to determine the up/down state of
the OSDs, the same output is used to determine the number of osd hosts in the
configuration. Prior to this change the determination was inferred through a
graphite query.
Paul Cuzner [Mon, 26 Jun 2017 05:07:50 +0000 (17:07 +1200)]
cephmetrics: Use a default value type for graphite, and assign default logging
Before this change, if a variable was defined in a class but NOT defined in it's attributes
the collector would fail. With this change a default of gauge is assigned.
In addition, a default logging level is set for all the collectors, if not specified by LogLevel
in the collectd.conf plugin
Paul Cuzner [Thu, 22 Jun 2017 21:16:08 +0000 (09:16 +1200)]
add home dashboard support
this change adds a _home_dashboard setting such that the grafana home dashboard
for the admin user can be changed to be the ceph-at-a-glance dashboard.
projects / cephmetrics.git / log