[Ganglia-developers] Gmetad custom RRAs

[Matthew C. <mat...@va...>]

After some consideration about how Gmetad handles the default and custom
RRAs, I think the system needs an overhaul.  The current .conf looks like
this:

===========================================================
# Format: 
# data_source "my cluster" [polling interval] address1:port addreses2:port
# 
# The keyword 'data_source' must immediately be followed by a unique
# string which identifies the source, then an optional polling interval in 
# seconds. The source will be polled at this interval on average. 
# If the polling interval is omitted, 15sec is asssumed. 
#
# A list of machines which service the data source follows, in the 
# format ip:port, or name:port. If a port is not specified then 8649
# (the default gmond port) is assumed.
# default: There is no default value
#
# data_source "my cluster" 10 localhost  my.machine.edu:8649  1.2.3.5:8655
# data_source "my grid" 50 1.3.4.7:8655 grid.org:8651 grid-backup.org:8651
# data_source "another source" 1.3.4.7:8655  1.3.4.8

#
# Round-Robin Archives
# You can specify custom Round-Robin archives here (defaults are listed
# below)
#
# RRAs 0 "RRA:AVERAGE:0.5:1:244" "RRA:AVERAGE:0.5:24:244" \
#        "RRA:AVERAGE:0.5:168:244" "RRA:AVERAGE:0.5:672:244" \
#        "RRA:AVERAGE:0.5:5760:374"#

#--------------------------------------------------------------------------
# The name of this Grid. All the data sources above will be wrapped in a
# GRID
# tag with this name.
# default: Unspecified
gridname "My Grid"

============================================================

As I see it, this configuration has two serious flaws:
- The gmetad grid (and thus the main __SummaryInfo__ RRDs) that is
associated with this gmetad.conf does not have a step interval associated
with it
- All data sources must either use the default RRAs or the custom RRAs,
which means that in order to work properly, the RRAs must be designed with
the lowest interval in mind

I think it should be changed to work something like this:

============================================================
# Data sources
# Format: 
# data_source [RRA id] <source name> [polling interval] <address1>[:port]
[address2[:port]] ...
# 
# Every data source is assigned to a set of RRAs. The default assignment
# is to id 0, which maps to a default set of RRAs that dynamically
# adjusts to the data source polling interval.
#
# The source name is required and must be unique.
#
# The optional polling interval in seconds affects the frequency at which
# the data source is polled for meta-data. It also affects the step size
# for the source's RRDs. If omitted, 15sec is assumed. 
#
# A list of machines which service the data source follows, in the 
# format ip:port, or name:port. If a port is not specified then 8649
# (the default gmond port) is assumed.  At least one IP or name
# must be listed.
#
# data_source "my cluster" 10 localhost  my.machine.edu:8649  1.2.3.5:8655
# data_source 2 "my grid" 50 1.3.4.7:8655 grid.org:8651 grid-backup.org:8651
# data_source 1 "another source" 1.3.4.7:8655  1.3.4.8

# Round-Robin Archives
# Format:
# RRAs <RRA id> <"first RRA definition"> ["second RRA definition"] .
#
# You can specify custom sets of Round-Robin archives here. Custom RRA sets
# are not dynamically adjusted based on the data source interval, so be
# careful when creating them.
#
# The default set is dynamically generated based on the data source
# interval, the five basic time periods (hour, day, week, month, year),
# and a reasonable amount of data averaging (i.e. oriented toward graphs
# that are around 250 pixels wide)
#
# For a 15 second interval, the default RRA set is:
# RRAs 0 "RRA:AVERAGE:0.5:1:244" "RRA:AVERAGE:0.5:24:244" \
#        "RRA:AVERAGE:0.5:168:244" "RRA:AVERAGE:0.5:672:244" \
#        "RRA:AVERAGE:0.5:5760:374"
#

#--------------------------------------------------------------------------
# The name of this Grid. All the data sources above will be wrapped in a
# GRID tag with this name.
# default: Unspecified
gridname "My Grid"

# A time interval in seconds that controls how often the grid's summary data
# is updated
# default: 15sec
gridstep 10

============================================================

So, what do you all think?  I think the current system hides the true
complexity of the RRD system behind inadequate documentation.  At the very
least, that part should improve. :)

Regards,
Matt Chambers