[checkmk-commits] RRD data query has configurable limit on data length

Module: check_mk Branch: master Commit: ab6113372ffa64ad4968374353db52c94c3d7d9d URL: http://git.mathias-kettner.de/git/?p=check_mk.git;a=commit;h=ab6113372ffa64… Author: Óscar Nájera <on(a)mathias-kettner.de> Date: Fri Apr 5 15:46:08 2019 +0200 RRD data query has configurable limit on data length RRDtool has a configurable default limit of 400 points per query. Livestatus did not configure that value until now, but we may need to do queries over large periods of time and we don't want down sampling. This change offers backwards compatibility, leaving this parameter as optional. The off by one in the query to result arises on the new way result bounds are given. Now the result bounds contain the query bounds instead of providing a shift of bounds between query and response. CMK-1935 Change-Id: I2bcd22396887fe04f7617cb6e1248225662482c1 --- cmk/utils/prediction.py | 8 +++----- tests/integration/cmk_base/test_services_prediction.py | 10 ++++++++++ 2 files changed, 13 insertions(+), 5 deletions(-) diff --git a/cmk/utils/prediction.py b/cmk/utils/prediction.py index 8aab80d..c1236af 100644 --- a/cmk/utils/prediction.py +++ b/cmk/utils/prediction.py @@ -111,15 +111,13 @@ class TimeSeries(object): return self.start == other.start and self.end == other.end and self.step == other.step and self.values == other.values -def get_rrd_data(hostname, service_description, varname, cf, fromtime, untiltime): +def get_rrd_data(hostname, service_description, varname, cf, fromtime, untiltime, max_entries=400): """Fetch RRD historic metrics data of a specific service, within the specified time range returns a TimeSeries object holding interval and data information Query to livestatus always returns if database is found, thus: - Values can be None when there is no data for a given timestamp - - Livestatus/rrdtool returns a maximum of 400 data-points per query, - even if better resolution is available within the time range. - Reply from livestatus/rrdtool is always enough to describe the queried interval. That means, the returned bounds are always outside the queried interval. @@ -140,11 +138,11 @@ def get_rrd_data(hostname, service_description, varname, cf, fromtime, untiltime rpn = "%s.%s" % (varname, cf.lower()) # "MAX" -> "max" lql = "GET services\n" \ - "Columns: rrddata:m1:%s:%s:%s:%s\n" \ + "Columns: rrddata:m1:%s:%s:%s:%s:%s\n" \ "OutputFormat: python\n" \ "Filter: host_name = %s\n" \ "Filter: description = %s\n" % tuple(map(livestatus.lqencode, - map(str, (rpn, fromtime, untiltime, step, + map(str, (rpn, fromtime, untiltime, step, max_entries, hostname, service_description)))) try: diff --git a/tests/integration/cmk_base/test_services_prediction.py b/tests/integration/cmk_base/test_services_prediction.py index fa6e331..a5a279b 100644 --- a/tests/integration/cmk_base/test_services_prediction.py +++ b/tests/integration/cmk_base/test_services_prediction.py @@ -88,6 +88,16 @@ def test_get_rrd_data(cfg_setup, utcdate, timezone, period, result): assert (timeseries.step, len(timeseries.values)) == result +(a)pytest.mark.parametrize("max_entriestries, result", [(400, (180, 401)), (20, (3600, 21)), + (50, (1800, 41)), (1000, (120, 600)), + (1200, (60, 1200))]) +def test_get_rrd_data_point_max(cfg_setup, max_entries, result): + from_time, until_time = 1543430040, 1543502040 + timeseries = cmk.utils.prediction.get_rrd_data('test-prediction', 'CPU load', 'load15', 'MAX', + from_time, until_time, max_entries) + assert (timeseries.step, len(timeseries.values)) == result + + @pytest.mark.parametrize('utcdate, timezone, params, reference', [ ("2018-09-01 07:00", "Europe/Berlin", { 'period': 'wday',