add update of checkpoint timestamps

This commit is contained in:
Christoph Kluge 2022-06-21 15:24:21 +02:00
parent f4ed659be2
commit dde48d2394

View File

@ -11,6 +11,7 @@ use Data::Dumper;
use Data::Walk; use Data::Walk;
use Scalar::Util qw( reftype ); use Scalar::Util qw( reftype );
use Time::Piece; use Time::Piece;
use Sort::Versions;
## NOTE: Based on Jan: migrateCC-jobArchive.pl ## NOTE: Based on Jan: migrateCC-jobArchive.pl
@ -54,11 +55,11 @@ my $targetDir = './cc-backend/var/job-archive';
my @Clusters; my @Clusters;
my $src = './data/job-archive'; my $src = './data/job-archive';
chomp($checkpointStart=`date --date 'TZ="Europe/Berlin" 0:00 7 days ago' +%s`); chomp(my $checkpointStart=`date --date 'TZ="Europe/Berlin" 0:00 7 days ago' +%s`);
my $halfday = 43200; my $halfday = 43200;
my $targetDirCheckpoints = './data/cc-metric-store_new' my $targetDirCheckpoints = './data/cc-metric-store_new';
my $srcCheckpoints = './data/cc-metric-store' my $srcCheckpoints = './data/cc-metric-store';
my @ClustersCheckpoints my @ClustersCheckpoints;
## Get Clusters ## Get Clusters
opendir my $dh, $src or die "can't open directory: $!"; opendir my $dh, $src or die "can't open directory: $!";
@ -81,7 +82,7 @@ while ( readdir $dhc ) {
# start for jobarchive # start for jobarchive
foreach my $cluster ( @Clusters ) { foreach my $cluster ( @Clusters ) {
print "Starting to update startTime for $cluster\n"; print "Starting to update startTime in job-archive for $cluster\n";
opendir my $dhLevel1, "$src/$cluster" or die "can't open directory: $!"; opendir my $dhLevel1, "$src/$cluster" or die "can't open directory: $!";
while ( readdir $dhLevel1 ) { while ( readdir $dhLevel1 ) {
@ -95,8 +96,8 @@ foreach my $cluster ( @Clusters ) {
my $level2 = $_; my $level2 = $_;
my $src = "$src/$cluster/$level1/$level2"; my $src = "$src/$cluster/$level1/$level2";
my $target = "$targetDir/$cluster/$level1/$level2/"; my $target = "$targetDir/$cluster/$level1/$level2/";
my $oldsrc = $src; my $oldsrc = $src;
if ( ! -e "$src/meta.json") { if ( ! -e "$src/meta.json") {
my @files = read_dir($src); my @files = read_dir($src);
if (!@files) { if (!@files) {
@ -109,8 +110,10 @@ foreach my $cluster ( @Clusters ) {
rmtree $oldsrc; rmtree $oldsrc;
next; next;
} }
my $str = read_file("$src/meta.json"); my $str = read_file("$src/meta.json");
my $json = decode_json($str); my $json = decode_json($str);
$FIRST = 1; $FIRST = 1;
walk \&process, $json; walk \&process, $json;
@ -125,9 +128,9 @@ foreach my $cluster ( @Clusters ) {
$target .= $json->{startTime}; $target .= $json->{startTime};
if ( not -d $target ){ if ( not -d $target ){
print "Writing files\n"; # print "Writing files\n";
print "$cluster/$level1/$level2\n"; # print "$cluster/$level1/$level2\n";
make_path($target); make_path($target);
$str = encode_json($json); $str = encode_json($json);
@ -142,5 +145,78 @@ foreach my $cluster ( @Clusters ) {
} }
} }
} }
print "Done for job-archive\n";
sleep(2);
print "Done\n"; # start for checkpoints
foreach my $cluster ( @ClustersCheckpoints ) {
print "Starting to update startTime in checkpoint-files for $cluster\n";
opendir my $dhLevel1, "$srcCheckpoints/$cluster" or die "can't open directory: $!";
while ( readdir $dhLevel1 ) {
chomp; next if $_ eq '.' or $_ eq '..';
my $level1 = $_;
if ( -d "$srcCheckpoints/$cluster/$level1" ) {
my $srcCheckpoints = "$srcCheckpoints/$cluster/$level1/";
my $target = "$targetDirCheckpoints/$cluster/$level1/";
my $oldsrc = $srcCheckpoints;
my @files;
if ( -e "$srcCheckpoints/1609459200.json") { # 1609459200 == First Checkpoint time in latest dump
@files = read_dir($srcCheckpoints);
my $length = @files;
if (!@files || $length != 14) { # needs 14 files == 7 days worth of data
next;
}
}
if ( ! -e "$srcCheckpoints/1609459200.json") {
# rmtree $oldsrc;
next;
}
my @sortedFiles = sort { versioncmp($a,$b) } @files; # sort alphanumerically: _Really_ start with index == 0 == 1609459200.json
while (my ($index, $file) = each(@sortedFiles)) {
# print "$file\n";
my $str = read_file("$srcCheckpoints/$file");
my $json = decode_json($str);
my $timestamp = $checkpointStart + ($index * $halfday);
my $oldTimestamp = $json->{from};
# print "$oldTimestamp -> $timestamp in $srcCheckpoints\n";
$json->{from} = $timestamp;
foreach my $metric (keys %{$json->{metrics}}) {
$json->{metrics}->{$metric}->{start} -= $oldTimestamp;
$json->{metrics}->{$metric}->{start} += $timestamp;
}
my $targetFile = "$target/$timestamp.json";
make_path($target);
$str = encode_json($json);
write_file("$targetFile", $str);
}
# if ( not -d $target ){
# print "Writing files\n";
#
# print "$cluster/$level1/$level2\n";
# make_path($target);
#
# $str = encode_json($json);
# write_file("$target/meta.json", $str);
#
# $str = read_file("$srcCheckpoints/data.json");
# write_file("$target/data.json", $str);
# } else {
# #rmtree $src;
# }
}
}
}
print "Done for checkpoints\n";