Description
Librato tracks and stores several different types of measurements as time-series data. Each measurement corresponds to a particular point in time. Queries are typically made to request the values over some time interval that is a subset of the entire series e.g. the last hour, last Monday, etc. The search parameters described below may be used in different combinations to specify a time interval when making queries against metrics.
Request Parameters
If an explicit interval (i.e. start_time to end_time) is specified, the
response contains all measurements that fall within the interval. In this
scenario the parameter start_time must be set, while end_time
may be set or left to default to the current time.
An interval can also be implicitly specified through a count
parameter. If count is set to N alongside with either (but
not both) start_time or end_time, the response covers the time
interval that covers N measurements. The value of end_time
always defaults to the current time, so a request for the last N
measurements only requires the count parameter set to N.
start_time The unix timestamp indicating the start time of the desired interval.
end_time The unix timestamp indicating the end time of the desired interval. If left unspecified it defaults to the current time.
count The number of measurements desired. When specified as N in conjunction with
start_time, the response contains the first N measurements afterstart_time. When specified as N in conjunction withend_time, the response contains the last N measurements beforeend_time.resolution A resolution for the response as measured in seconds. If the original measurements were reported at a higher resolution than specified in the request, the response contains averaged measurements.
Pagination
If a request does not include an explicit count and the matched
data range includes more points than the maximum return size, then
the response will include a pagination hint. In this case the
response will include a parameter next_time in the query
top-level response section. The next_time will be set to the
epoch second start time of the next matching element of the
original request. This hint serves two purposes: 1) it indicates
there are more matching elements, but that the original request
was truncated due to the response limit, and 2) it provides the
start_time value that should be used in a subsequent request. If
the original request is resubmitted with the start_time set to
the returned next_time the response will include the next set of
matching elements in the set. It may require multiple requests to
page through the entire set if the number of matching elements is
large.
For example, if the response includes the following query
section it implies there are more points and that start_time
should be set to 1305562061 to retrieve the next matching
elements.
"query" : {
"next_time" : 1305562061
}
Examples
TBD