Skip to content

Commit

Permalink
Calculating begin/end needed further refinement
Browse files Browse the repository at this point in the history 
- See code comments for details
  • Loading branch information
@atheurer
atheurer committed Oct 31, 2024
1 parent 49ff11e commit 81582e7
Showing 1 changed file with 86 additions and 37 deletions.
123 changes: 86 additions & 37 deletions rickshaw-index
View file
Original file line number Diff line number Diff line change
Expand Up @@ -376,6 +376,8 @@ sub index_metrics {
my $num_metric_docs_submitted = 0;
my $earliest_begin;
my $latest_end;
my $pri_earliest_begin;
my $pri_latest_end;
my $coder = JSON::XS->new->canonical;

my $dir = pushd($metr_dir);
Expand Down Expand Up @@ -545,15 +547,21 @@ sub index_metrics {
$ndjson = "";
$count = 0;
}
if (defined $primary_metric) {
if ($type{$idx} eq $primary_metric and $source{$idx} eq $benchmark) {
$primary_metric_found = 1;
if (not defined $earliest_begin or $earliest_begin > $metr_data_doc{'metric_data'}{'begin'}) {
$earliest_begin = $metr_data_doc{'metric_data'}{'begin'};
}
if (not defined $latest_end or $latest_end < $metr_data_doc{'metric_data'}{'end'}) {
$latest_end = $metr_data_doc{'metric_data'}{'end'};
}

if (defined $primary_metric and $type{$idx} eq $primary_metric and $source{$idx} eq $benchmark) {
$primary_metric_found = 1;
if (not defined $pri_earliest_begin or $pri_earliest_begin > $metr_data_doc{'metric_data'}{'begin'}) {
$pri_earliest_begin = $metr_data_doc{'metric_data'}{'begin'};
}
if (not defined $pri_latest_end or $pri_latest_end < $metr_data_doc{'metric_data'}{'end'}) {
$pri_latest_end = $metr_data_doc{'metric_data'}{'end'};
}
} else {
if (not defined $earliest_begin or $earliest_begin > $metr_data_doc{'metric_data'}{'begin'}) {
$earliest_begin = $metr_data_doc{'metric_data'}{'begin'};
}
if (not defined $latest_end or $latest_end < $metr_data_doc{'metric_data'}{'end'}) {
$latest_end = $metr_data_doc{'metric_data'}{'end'};
}
}
}
Expand All @@ -575,14 +583,14 @@ sub index_metrics {
}

if (defined $primary_metric and $primary_metric_found == 1) {
if (defined $earliest_begin and defined $latest_end) {
return ($num_metric_docs_submitted, $earliest_begin, $latest_end);
if (defined $pri_earliest_begin and defined $pri_latest_end) {
return ($num_metric_docs_submitted, 1, $pri_earliest_begin, $pri_latest_end);
} else {
printf "ERROR: index_metrics() primary_metric found, but undefined earliest_begin and/or undefined latest_end, exiting\n";
exit 1;
}
} else {
return $num_metric_docs_submitted;
return ($num_metric_docs_submitted, 0, $earliest_begin, $latest_end);
}
}

Expand Down Expand Up @@ -1067,20 +1075,53 @@ if (exists $result{'iterations'}) {
# while indexing the metrics.
my $earliest_begin;
my $latest_end;
my $pm_earliest_begin;
my $pm_latest_end;
my $base_metric_doc_ref = create_es_doc("metric_desc", $iter_idx, $sample_idx, $period_idx);
print "period: $$this_sample{'periods'}[$period_idx]{'name'}\n";
my $primary_metric_found = 0;
for (my $j = 0; $j < scalar(@{ $data{'periods'}[$k]{'metric-files'} }); $j++) {
# Metric data is still in other file(s). For each member in 'metric-files' array,
# there should be a 2 files with the same prefix
my $metric_file_prefix = $data{'periods'}[$k]{'metric-files'}[$j];
my $metric_dir = $run_dir . "/" . $cs_id_dir;
my $this_begin;
my $this_end;
# index_metric() to return the earliest-begin and latest-end for metric types matching the primary-metric
(my $num_metric_docs_submitted, $this_begin, $this_end) = index_metrics('queue', $metric_dir, $metric_file_prefix, $cs_name, $cs_id, $base_metric_doc_ref, $data{'benchmark'}, $data{'primary-metric'});
# From processing all metric files, get the very-earliest-begin and very-latest-end
# This is to ensure we get the biggest time range for a *specific* client
if (defined $this_begin and defined $this_end) {
my $this_pm_found;
# index_metric(): Note that if a primary metric is found on this data,
# it returns the earliest begin and latest end for only metric data from the
# primary metric. If the primary metric is not found, then it returns the
# earliest begin and latest end for all metric data found.
#
# Given that there can be multiple metric data files to process, and it's
# possible that one data file could contain primary metric data, but another
# file could contain no primary metric data, we have to be prepared to later use
# the begin/end from either all the metric data or just the primary metric data.
#
# After all files are processed, if there is no data for the primary metric found,
# then the earliest begin and latest end from *all* metric data can be used.
# However, if at least one metric data file does contains primary metric data,
# then only the earliest begin and latest end for the primary metric can be used.
#
# What exactly are these earliest begin and latest end used for? To determine the
# period's begin and end, including facotring for multiple clients, further below.

(my $num_metric_docs_submitted, $this_pm_found, $this_begin, $this_end) =
index_metrics('queue', $metric_dir, $metric_file_prefix,
$cs_name, $cs_id, $base_metric_doc_ref,
$data{'benchmark'}, $data{'primary-metric'});
# From processing all metric files (for this-client in this-priod),
# get the very-earliest begin and very-latest end

if ($this_pm_found) {
$primary_metric_found = 1;
if (not defined $pm_earliest_begin or $pm_earliest_begin > $this_begin) {
$pm_earliest_begin = $this_begin;
}
if (not defined $pm_latest_end or $pm_latest_end < $this_end) {
$pm_latest_end = $this_end;
}
} else {
if (not defined $earliest_begin or $earliest_begin > $this_begin) {
$earliest_begin = $this_begin;
}
Expand All @@ -1089,30 +1130,38 @@ if (exists $result{'iterations'}) {
}
}
}
if ($$this_sample{'periods'}[$period_idx]{'name'} eq $data{'primary-period'}) {
if (not defined $earliest_begin or not defined $latest_end) {
if ($primary_metric_found) {
if (not defined $pm_earliest_begin or not defined $pm_latest_end) {
print "Either earliest_begin and/or latest_end were not defined, exiting";
exit 1;
} else {
$earliest_begin = $pm_earliest_begin;
$latest_end = $pm_latest_end;
}
# Now if this client/server's earliest_begin is *later* than a defined begin for the consolidated period,
# we need to adjust the begin for the consolidated period to match this client/server's earliest_begin.
# This ensures the consolidated period always has samples from every single client/server for the entire
# period.
if (not defined $$this_sample{'periods'}[$period_idx]{'begin'} or $$this_sample{'periods'}[$period_idx]{'begin'} < $earliest_begin) {
$$this_sample{'periods'}[$period_idx]{'begin'} = $earliest_begin;
debug_log(sprintf "client/server's ID %d begin is after current sample begin, so assigning sample begin to %d\n", $cs_id, $earliest_begin);
}
if (not defined $$this_sample{'periods'}[$period_idx]{'end'} or $$this_sample{'periods'}[$period_idx]{'end'} > $latest_end) {
$$this_sample{'periods'}[$period_idx]{'end'} = $latest_end;
debug_log(sprintf "client/server's ID %d end is before current sample end, so assigning sample begin to %d\n", $cs_id, $latest_end);
}
if (! defined $result{'run.begin'} or $result{'begin'} > $$this_sample{'periods'}[$period_idx]{'begin'}) {
$result{'begin'} = $$this_sample{'periods'}[$period_idx]{'begin'};
}
if (! defined $result{'end'} or $result{'end'} < $$this_sample{'periods'}[$period_idx]{'end'}) {
$result{'end'} = $$this_sample{'periods'}[$period_idx]{'end'};
}
}
if (not defined $earliest_begin or not defined $latest_end) {
print "Either earliest_begin and/or latest_end were not defined, exiting";
exit 1;
}
# Now if this client/server's earliest_begin is *later* than a defined begin for the consolidated period,
# we need to adjust the begin for the consolidated period to match this client/server's earliest_begin.
# This ensures the consolidated period always has samples from every single client/server for the entire
# period.
if (not defined $$this_sample{'periods'}[$period_idx]{'begin'} or $$this_sample{'periods'}[$period_idx]{'begin'} < $earliest_begin) {
$$this_sample{'periods'}[$period_idx]{'begin'} = $earliest_begin;
debug_log(sprintf "client/server's ID %d begin is after current sample begin, so assigning sample begin to %d\n", $cs_id, $earliest_begin);
}
if (not defined $$this_sample{'periods'}[$period_idx]{'end'} or $$this_sample{'periods'}[$period_idx]{'end'} > $latest_end) {
$$this_sample{'periods'}[$period_idx]{'end'} = $latest_end;
debug_log(sprintf "client/server's ID %d end is before current sample end, so assigning sample begin to %d\n", $cs_id, $latest_end);
}
if (! defined $result{'run.begin'} or $result{'begin'} > $$this_sample{'periods'}[$period_idx]{'begin'}) {
$result{'begin'} = $$this_sample{'periods'}[$period_idx]{'begin'};
}
if (! defined $result{'end'} or $result{'end'} < $$this_sample{'periods'}[$period_idx]{'end'}) {
$result{'end'} = $$this_sample{'periods'}[$period_idx]{'end'};
}

queue_es_doc("period", $run_dir . "/" . $this_samp_dir, $iter_idx, $sample_idx, $period_idx);
}
}
Expand Down

0 comments on commit 81582e7

Please sign in to comment.