mirror of
https://git.savannah.gnu.org/git/parallel.git
synced 2024-11-22 14:07:55 +00:00
parallel: Saving to SQL is now done using sql->output().
This commit is contained in:
parent
0c41e2d2df
commit
e305356898
76
src/parallel
76
src/parallel
|
@ -64,7 +64,7 @@ if($opt::sqlmaster) {
|
|||
$Global::sql->create_table($#input_source_fh+1);
|
||||
if($opt::sqlworker) {
|
||||
# Start a real --sqlworker in the background later
|
||||
$Global::sqlworker = 1;
|
||||
$Global::start_sqlworker = 1;
|
||||
$opt::sqlworker = undef;
|
||||
}
|
||||
}
|
||||
|
@ -1223,7 +1223,7 @@ sub check_invalid_option_combinations {
|
|||
|
||||
sub init_globals {
|
||||
# Defaults:
|
||||
$Global::version = 20161123;
|
||||
$Global::version = 20161208;
|
||||
$Global::progname = 'parallel';
|
||||
$Global::infinity = 2**31;
|
||||
$Global::debug = 0;
|
||||
|
@ -2460,9 +2460,9 @@ sub drain_job_queue {
|
|||
while($opt::sqlmaster and not $Global::sql->finished()) {
|
||||
# SQL master
|
||||
$sleep = ::reap_usleep($sleep);
|
||||
if($Global::sqlworker) {
|
||||
if($Global::start_sqlworker) {
|
||||
# Start an SQL worker as we are now sure there is work to do
|
||||
$Global::sqlworker = 0;
|
||||
$Global::start_sqlworker = 0;
|
||||
if(fork()) {
|
||||
# skip
|
||||
} else {
|
||||
|
@ -8229,6 +8229,10 @@ sub linebuffer_print {
|
|||
}
|
||||
# Print up to and including the last \n
|
||||
print $out_fd substr($$partial,0,$i);
|
||||
if($opt::sqlworker) {
|
||||
push (@{$self->{'sqlworker'}{$fdno}},
|
||||
substr($$partial,0,$i));
|
||||
}
|
||||
# Remove the printed part
|
||||
substr($$partial,0,$i) = "";
|
||||
}
|
||||
|
@ -8247,6 +8251,9 @@ sub linebuffer_print {
|
|||
$$partial =~ s/^/$tag/gm;
|
||||
}
|
||||
print $out_fd $$partial;
|
||||
if($opt::sqlworker) {
|
||||
push (@{$self->{'sqlworker'}{$fdno}},$$partial);
|
||||
}
|
||||
}
|
||||
# Release the memory
|
||||
undef $$partial;
|
||||
|
@ -8256,24 +8263,9 @@ sub linebuffer_print {
|
|||
# decompress done:
|
||||
# copy to sql (if needed)
|
||||
# then close fh
|
||||
if($opt::sqlworker) {
|
||||
my @output;
|
||||
my $tag = $self->tag();
|
||||
seek $in_fh, 0, 0;
|
||||
while(<$in_fh>) {
|
||||
push @output, $tag,$_;
|
||||
}
|
||||
if($fdno == 1) {
|
||||
if(not $opt::results) {
|
||||
$Global::sql->update("SET Stdout = ? WHERE Seq = ".$self->seq(),
|
||||
join("",@output));
|
||||
}
|
||||
} else {
|
||||
if(not $opt::results) {
|
||||
$Global::sql->update("SET Stderr = ? WHERE Seq = ".$self->seq(),
|
||||
join("",@output));
|
||||
}
|
||||
}
|
||||
if($opt::sqlworker and not $opt::results) {
|
||||
$Global::sql->output($fdno, $self->seq(),
|
||||
join("",@{$self->{'sqlworker'}{$fdno}}));
|
||||
}
|
||||
close $in_fh;
|
||||
if($? and $opt::compress) {
|
||||
|
@ -8308,15 +8300,9 @@ sub tag_print {
|
|||
}
|
||||
if($fdno == 1) {
|
||||
$self->add_returnsize($outputlength);
|
||||
if($opt::sqlworker and not $opt::results) {
|
||||
$Global::sql->update("SET Stdout = ? WHERE Seq = ".$self->seq(),
|
||||
join("",@output));
|
||||
}
|
||||
} else {
|
||||
if($opt::sqlworker and not $opt::results) {
|
||||
$Global::sql->update("SET Stderr = ? WHERE Seq = ".$self->seq(),
|
||||
join("",@output));
|
||||
}
|
||||
}
|
||||
if($opt::sqlworker and not $opt::results) {
|
||||
$Global::sql->output($fdno, $self->seq(), join("",@output));
|
||||
}
|
||||
close $in_fh;
|
||||
if($? and $opt::compress) {
|
||||
|
@ -8347,15 +8333,10 @@ sub normal_print {
|
|||
}
|
||||
if($fdno == 1) {
|
||||
$self->add_returnsize($outputlength);
|
||||
if($opt::sqlworker and not $opt::results) {
|
||||
$Global::sql->update("SET Stdout = ? WHERE Seq = ".$self->seq(),
|
||||
join("",@output));
|
||||
}
|
||||
} else {
|
||||
if($opt::sqlworker and not $opt::results) {
|
||||
$Global::sql->update("SET Stderr = ? WHERE Seq = ".$self->seq(),
|
||||
join("",@output));
|
||||
}
|
||||
}
|
||||
if($opt::sqlworker and not $opt::results) {
|
||||
$Global::sql->output($fdno, $self->seq(),
|
||||
join("",@output));
|
||||
}
|
||||
close $in_fh;
|
||||
if($? and $opt::compress) {
|
||||
|
@ -10222,7 +10203,7 @@ sub get_alias {
|
|||
if($dburl) {
|
||||
return get_alias($dburl.$rest);
|
||||
} else {
|
||||
Usage("$alias is not defined in @search");
|
||||
::error("$alias is not defined in @search");
|
||||
exit(-1);
|
||||
}
|
||||
}
|
||||
|
@ -10406,6 +10387,17 @@ sub update {
|
|||
$self->run("UPDATE $table $stmt",@_);
|
||||
}
|
||||
|
||||
sub output {
|
||||
my $self = shift;
|
||||
my $fdno = shift;
|
||||
my $seq = shift;
|
||||
my @output = @_;
|
||||
|
||||
my %column_of_fdno = ( 1 => "Stdout", 2 => "Stderr" );
|
||||
my $column = $column_of_fdno{$fdno};
|
||||
$self->update("SET $column = ? WHERE Seq = ".$seq, join("",@output));
|
||||
}
|
||||
|
||||
sub max_number_of_args {
|
||||
# Maximal number of args for this table
|
||||
my $self = shift;
|
||||
|
@ -10526,7 +10518,7 @@ sub finished {
|
|||
# Check if there are any jobs left in the SQL table that do not
|
||||
# have a "real" exitval
|
||||
my $self = shift;
|
||||
if($opt::wait) {
|
||||
if($opt::wait or $Global::start_sqlworker) {
|
||||
my $table = $self->table();
|
||||
my $rv = $self->get("select Seq,Exitval from $table where Exitval <= -1000 limit 1");
|
||||
return not $rv->[0];
|
||||
|
|
|
@ -1925,6 +1925,9 @@ Execute jobs via SQL server. Read the input sources variables from the
|
|||
table pointed to by I<DBURL>. The I<command> on the command line
|
||||
should be the same as given by B<--sqlmaster>.
|
||||
|
||||
If you have more than one B<--sqlworker> jobs may be run more than
|
||||
once.
|
||||
|
||||
|
||||
=item B<--ssh> I<sshcommand>
|
||||
|
||||
|
|
|
@ -76,6 +76,23 @@ where there is no easy way to convert an environment into commands
|
|||
that generate the environment.
|
||||
|
||||
|
||||
=head3 env_parallel.*
|
||||
|
||||
These are the files that implements the alias or function
|
||||
B<env_parallel> for a given shell. It could be argued that these
|
||||
should be put in some obscure place under /usr/lib, but by putting
|
||||
them in your path it becomes trivial to find the path to them and
|
||||
B<source> them:
|
||||
|
||||
source `which env_parallel.foo`
|
||||
|
||||
The beauty is that they can be put anywhere in the path without the
|
||||
user having to know the location. So if the user's path includes
|
||||
/afs/bin/i386_fc5 or /usr/pkg/parallel/bin or
|
||||
/usr/local/parallel/20161222/sunos5.6/bin the files can be put in the
|
||||
dir that makes most sense for the sysadmin.
|
||||
|
||||
|
||||
=head3 env_parallel.bash / env_parallel.zsh / env_parallel.ksh / env_parallel.pdksh
|
||||
|
||||
B<env_parallel.(bash|ksh|pdksh|zsh)> sets the function B<env_parallel>. It uses
|
||||
|
@ -731,6 +748,14 @@ commands (that cannot be parallelized) on each server, but run the
|
|||
same sequence on multiple servers.
|
||||
|
||||
|
||||
=head2 --shuf
|
||||
|
||||
When using B<--shuf> to shuffle the jobs, all jobs are read, then they
|
||||
are shuffled, and finally executed. When using SQL this makes the
|
||||
B<--sqlmaster> be the part that shuffles the jobs. The B<--sqlworker>s
|
||||
simply executes according to Seq number.
|
||||
|
||||
|
||||
=head2 Buffering on disk
|
||||
|
||||
GNU B<parallel> buffers on disk in $TMPDIR using files, that are
|
||||
|
|
Loading…
Reference in a new issue