parallel: stdin_buffer changed to block.

This commit is contained in:
Ole Tange 2015-11-22 11:12:47 +01:00
parent 5be0293a96
commit fc8e346bee
6 changed files with 124 additions and 107 deletions

50
NEWS
View file

@ -1,8 +1,8 @@
20151022
* --plus makes it possible to use {##} as a short had for {=
$_=$Global::JobQueue->total_jobs() =} which gives the the number of
jobs to run.
* --plus makes it possible to use {##} as a shorthand for
{= $_=$Global::JobQueue->total_jobs() =} which gives the the number
of jobs to run in total.
* {= $_=$Global::JobQueue->total_jobs() =} is incompatible with -X,
-m, and --xargs.
@ -37,30 +37,48 @@
20150922
* GNU Parallel was cited in: Flexible Modeling of Epidemics with an Empirical Bayes Framework http://journals.plos.org/ploscompbiol/article?id=10.1371%2Fjournal.pcbi.1004382
* GNU Parallel was cited in: Flexible Modeling of Epidemics with an
Empirical Bayes Framework
http://journals.plos.org/ploscompbiol/article?id=10.1371%2Fjournal.pcbi.1004382
* GNU Parallel was cited in: BL1: 2D Potts Model with a Twist https://sucs.swan.ac.uk/~rjames93/Dissertation.pdf
* GNU Parallel was cited in: BL1: 2D Potts Model with a Twist
https://sucs.swan.ac.uk/~rjames93/Dissertation.pdf
* GNU Parallel was cited in: DockBench: An Integrated Informatic Platform Bridging the Gap between the Robust Validation of Docking Protocols and Virtual Screening Simulations http://www.mdpi.com/1420-3049/20/6/9977/pdf
* GNU Parallel was cited in: DockBench: An Integrated Informatic
Platform Bridging the Gap between the Robust Validation of Docking
Protocols and Virtual Screening Simulations
http://www.mdpi.com/1420-3049/20/6/9977/pdf
* GNU Parallel was cited in: A Scalable Parallel Implementation of Evolutionary Algorithms for
Multi-Objective Optimization on GPUs http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7257074
* GNU Parallel was cited in: A Scalable Parallel Implementation of
Evolutionary Algorithms for Multi-Objective Optimization on GPUs
http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7257074
* GNU Parallel was cited in: Tools and techniques for computational reproducibility http://biorxiv.org/content/biorxiv/early/2015/07/17/022707.full.pdf
* GNU Parallel was cited in: Tools and techniques for computational
reproducibility
http://biorxiv.org/content/biorxiv/early/2015/07/17/022707.full.pdf
* GNU Parallel was cited in: How Can We Measure the Similarity Between Résumés of Selected Candidates for a Job? http://www.researchgate.net/publication/275954089_How_can_we_measure_the_similarity_between_rsums_of_selected_candidates_for_a_job
* GNU Parallel was cited in: How Can We Measure the Similarity Between
Résumés of Selected Candidates for a Job?
http://www.researchgate.net/publication/275954089_How_can_we_measure_the_similarity_between_rsums_of_selected_candidates_for_a_job
* GNU Parallel was cited in: Interplay of cell dynamics and epithelial tension during morphogenesis of the Drosophila pupal wing http://www.researchgate.net/profile/Raphael_Etournay/publication/279061859_Interplay_of_cell_dynamics_and_epithelial_tension_during_morphogenesis_of_the_Drosophila_pupal_wing/links/558a95ad08aeae8413bcceea.pdf
* GNU Parallel was cited in: Interplay of cell dynamics and epithelial
tension during morphogenesis of the Drosophila pupal wing
http://www.researchgate.net/profile/Raphael_Etournay/publication/279061859_Interplay_of_cell_dynamics_and_epithelial_tension_during_morphogenesis_of_the_Drosophila_pupal_wing/links/558a95ad08aeae8413bcceea.pdf
* Third-party selling GNU Parallel T-shirts http://www.aliexpress.com/item/2015F-BSO-GNU-LOGO-GNU-PARALLEL-men-s-shirt-sleeve-visual-illusion-error/32464827966.html
* Third-party selling GNU Parallel T-shirts
http://www.aliexpress.com/item/2015F-BSO-GNU-LOGO-GNU-PARALLEL-men-s-shirt-sleeve-visual-illusion-error/32464827966.html
* Joys of gnu parallel http://scottolesen.com/index.php/2015/08/26/joys-of-gnu-parallel/
* Joys of gnu parallel
http://scottolesen.com/index.php/2015/08/26/joys-of-gnu-parallel/
* Crop and resize images with bash and ImageMagick https://www.simonholywell.com/post/2015/08/image-resize-crop-bash-imagemagick/
* Crop and resize images with bash and ImageMagick
https://www.simonholywell.com/post/2015/08/image-resize-crop-bash-imagemagick/
* Three Ways to Script Processes in Parallel http://www.codeword.xyz/2015/09/02/three-ways-to-script-processes-in-parallel/
* Three Ways to Script Processes in Parallel
http://www.codeword.xyz/2015/09/02/three-ways-to-script-processes-in-parallel/
* What It Looks Like to Process 3.5 Million Books http://blog.gdeltproject.org/what-it-looks-like-to-process-3-5-million-books/
* What It Looks Like to Process 3.5 Million Books
http://blog.gdeltproject.org/what-it-looks-like-to-process-3-5-million-books/
* Bug fixes and man page updates.

View file

@ -4,7 +4,7 @@
Check that documentation is updated (compare to web):
Fixet for 20150922
Fixet for 20151022
git diff last-release-commit
Unmodified beta since last version => production
Unmodified alpha since last version => beta
@ -212,17 +212,15 @@ cc:Tim Cuthbertson <tim3d.junk@gmail.com>,
Ryoichiro Suzuki <ryoichiro.suzuki@gmail.com>,
Jesse Alama <jesse.alama@gmail.com>
Subject: GNU Parallel 20151022 ('Liquid Water / 9N314M') released <<[stable]>>
Subject: GNU Parallel 20151122 ('Kronan/Sharm el-Sheik/Bataclan') released <<[stable]>>
GNU Parallel 20151022 ('Liquid Water') <<[stable]>> has been released. It is available for download at: http://ftp.gnu.org/gnu/parallel/
GNU Parallel 20151122 ('Kronan/Sharm el-Sheik/Bataclan') <<[stable]>> has been released. It is available for download at: http://ftp.gnu.org/gnu/parallel/
<<No new functionality was introduced so this is a good candidate for a stable release.>>
Haiku of the month:
In parallel land
everything is quick and fast
Use GNU Parallel.
<<>>
-- Ole Tange
New in this release:
@ -236,25 +234,15 @@ New in this release:
* <<kontaktet 2015-06-22 Afventer svar fra journal>> GNU Parallel was used (unfortunately with wrong citation) in: TADSim: Discrete Event-Based Performance Prediction for Temperature-Accelerated Dynamics http://vruehle.de/publications/2015c.pdf
http://www.researchgate.net/profile/Christoph_Junghans/publication/276178326_TADSim_Discrete_Event-Based_Performance_Prediction_for_Temperature-Accelerated_Dynamics/links/55562b6708ae980ca60c8369.pdf
* --plus makes it possible to use {##} as a short had for {= $_=$Global::JobQueue->total_jobs() =} which gives the the number of jobs to run.
* << Update forventet juni Rachel har lige svaret >> GNU Parallel was used in: SISRS: Site Identification from Short Read Sequences https://github.com/rachelss/SISRS/
* GNU Parallel packaged for CERN CentOS: http://linuxsoft.cern.ch/cern/centos/7/cern/x86_64/repoview/parallel.html
* {= $_=$Global::JobQueue->total_jobs() =} is incompatible with -X, -m, and --xargs.
* Automating large numbers of tasks https://rcc.uchicago.edu/docs/tutorials/kicp-tutorials/running-jobs.html
* GNU Parallel is now mostly compatible with lsh (http://www.lysator.liu.se/~nisse/lsh/) and somewhat compatible with autossh (http://www.harding.motd.ca/autossh/).
* Max out your IOPs with GNU Parallel http://blog.bitratchet.com/2015/11/11/max-out-your-iops-with-gnu-parallel/
* --workdir ... now also works when run locally.
* GNU Parallel was cited in: There is no (75, 32, 10, 16) strongly regular graph http://arxiv.org/pdf/1509.05933.pdf
* GNU Parallel was cited in: Roary: rapid large-scale prokaryote pan genome analysis http://bioinformatics.oxfordjournals.org/content/early/2015/08/05/bioinformatics.btv421.full.pdf+html
* GNU Parallel is used in TraitAR: https://testpypi.python.org/pypi/traitar/0.1.4
* GNU Parallel is used in youtube-dl-parallel: https://github.com/dlh/youtube-dl-parallel
* A parallel and fast way to download multiple files http://onetipperday.blogspot.com/2015/10/a-parallel-and-fast-way-to-download.html
* Usar GNU Parallel para aumentar el rendimiento de tus scripts http://adrianarroyocalle.github.io/blog/2015/10/20/usar-gnu-parallel/
* GNU Parallel was cited in: Evolution of movement strategies under competitive interactions http://digital.csic.es/bitstream/10261/115973/1/evolution_movement_strategies_Kiziridis.pdf
* Bug fixes and man page updates.

View file

@ -435,8 +435,8 @@ sub spreadstdin {
my $something_written = 0;
for my $pid (keys %incomplete_jobs) {
my $job = $incomplete_jobs{$pid};
if($job->stdin_buffer_length()) {
$something_written += $job->non_block_write();
if($job->block_length()) {
$something_written += $job->non_blocking_write();
} else {
delete $incomplete_jobs{$pid}
}
@ -509,7 +509,7 @@ sub nindex {
# %Global::running
# Returns:
# $something_written = amount of bytes written
my ($header_ref,$block_ref,$recstart,$recend,$endpos) = @_;
my ($header_ref,$buffer_ref,$recstart,$recend,$endpos) = @_;
my $something_written = 0;
my $block_passed = 0;
my $sleep = 1;
@ -522,13 +522,13 @@ sub nindex {
values %Global::running);
}
while(my $job = shift @robin_queue) {
if($job->stdin_buffer_length() > 0) {
$something_written += $job->non_block_write();
if($job->block_length() > 0) {
$something_written += $job->non_blocking_write();
} else {
$job->set_stdin_buffer($header_ref,$block_ref,$endpos,$recstart,$recend);
$job->set_block($header_ref,$buffer_ref,$endpos,$recstart,$recend);
$block_passed = 1;
$job->set_virgin(0);
$something_written += $job->non_block_write();
$something_written += $job->non_blocking_write();
last;
}
}
@ -633,21 +633,21 @@ sub write_record_to_pipe {
# Input:
# $chunk_number = sequence number - to see if already run
# $header_ref = reference to header string to prepend
# $record_ref = reference to record to write
# $buffer_ref = reference to record to write
# $recstart = start string of record
# $recend = end string of record
# $endpos = position in $record_ref where record ends
# $endpos = position in $buffer_ref where record ends
# Uses:
# $Global::job_already_run
# $opt::roundrobin
# @Global::virgin_jobs
# Returns:
# Number of chunks written (0 or 1)
my ($chunk_number,$header_ref,$record_ref,$recstart,$recend,$endpos) = @_;
my ($chunk_number,$header_ref,$buffer_ref,$recstart,$recend,$endpos) = @_;
if($endpos == 0) { return 0; }
if(vec($Global::job_already_run,$chunk_number,1)) { return 1; }
if($opt::roundrobin) {
return round_robin_write($header_ref,$record_ref,$recstart,$recend,$endpos);
return round_robin_write($header_ref,$buffer_ref,$recstart,$recend,$endpos);
}
# If no virgin found, backoff
my $sleep = 0.0001; # 0.01 ms - better performance on highend
@ -655,28 +655,31 @@ sub write_record_to_pipe {
::debug("pipe", "No virgin jobs");
$sleep = ::reap_usleep($sleep);
# Jobs may not be started because of loadavg
# or too little time between each ssh login.
# or too little time between each ssh login
# or retrying failed jobs.
start_more_jobs();
}
my $job = shift @Global::virgin_jobs;
# Job is no longer virgin
$job->set_virgin(0);
# We ignore the removed rec_sep which is technically wrong.
$job->add_transfersize($endpos + length $$header_ref);
if(fork()) {
# Skip
} else {
# Chop of at $endpos as we do not know how many rec_sep will
# be removed.
substr($$record_ref,$endpos,length $$record_ref) = "";
# Remove rec_sep
if($opt::remove_rec_sep) {
Job::remove_rec_sep($record_ref,$recstart,$recend);
if(1) {
# We ignore the removed rec_sep which is technically wrong.
$job->add_transfersize($endpos + length $$header_ref);
if(fork()) {
# Skip
} else {
# Chop of at $endpos as we do not know how many rec_sep will
# be removed.
substr($$buffer_ref,$endpos,length $$buffer_ref) = "";
# Remove rec_sep
if($opt::remove_rec_sep) {
Job::remove_rec_sep($buffer_ref,$recstart,$recend);
}
$job->write($header_ref);
$job->write($buffer_ref);
close $job->fh(0,"w");
exit(0);
}
$job->write($header_ref);
$job->write($record_ref);
close $job->fh(0,"w");
exit(0);
}
close $job->fh(0,"w");
return 1;
@ -1100,7 +1103,7 @@ sub parse_options {
sub init_globals {
# Defaults:
$Global::version = 20151022;
$Global::version = 20151023;
$Global::progname = 'parallel';
$Global::infinity = 2**31;
$Global::debug = 0;
@ -6198,32 +6201,38 @@ sub write {
}
}
sub set_stdin_buffer {
sub set_block {
# Copy stdin buffer from $block_ref up to $endpos
# Prepend with $header_ref
# Prepend with $header_ref if virgin (i.e. not --roundrobin)
# Remove $recstart and $recend if needed
# Input:
# $header_ref = ref to $header to prepend
# $block_ref = ref to $block to pass on
# $buffer_ref = ref to $buffer containing the block
# $endpos = length of $block to pass on
# $recstart = --recstart regexp
# $recend = --recend regexp
# Returns:
# N/A
my $self = shift;
my ($header_ref,$block_ref,$endpos,$recstart,$recend) = @_;
$self->{'stdin_buffer'} = ($self->virgin() ? $$header_ref : "").substr($$block_ref,0,$endpos);
my ($header_ref,$buffer_ref,$endpos,$recstart,$recend) = @_;
$self->{'block'} = ($self->virgin() ? $$header_ref : "").substr($$buffer_ref,0,$endpos);
if($opt::remove_rec_sep) {
remove_rec_sep(\$self->{'stdin_buffer'},$recstart,$recend);
remove_rec_sep(\$self->{'block'},$recstart,$recend);
}
$self->{'stdin_buffer_length'} = length $self->{'stdin_buffer'};
$self->{'stdin_buffer_pos'} = 0;
$self->add_transfersize($self->{'stdin_buffer_length'});
$self->{'block_length'} = length $self->{'block'};
$self->{'block_pos'} = 0;
$self->add_transfersize($self->{'block_length'});
}
sub stdin_buffer_length {
sub block_ref {
my $self = shift;
return $self->{'stdin_buffer_length'};
return \$self->{'block'};
}
sub block_length {
my $self = shift;
return $self->{'block_length'};
}
sub remove_rec_sep {
@ -6234,26 +6243,26 @@ sub remove_rec_sep {
$$block_ref =~ s/$recend$//os;
}
sub non_block_write {
sub non_blocking_write {
my $self = shift;
my $something_written = 0;
use POSIX qw(:errno_h);
# for loop used to avoid copying substr: $buf will be an alias for the substr
for my $buf (substr($self->{'stdin_buffer'},$self->{'stdin_buffer_pos'})) {
for my $buf (substr($self->{'block'},$self->{'block_pos'})) {
my $in = $self->fh(0,"w");
my $rv = syswrite($in, $buf);
if (!defined($rv) && $! == EAGAIN) {
# would block
$something_written = 0;
} elsif ($self->{'stdin_buffer_pos'}+$rv != $self->{'stdin_buffer_length'}) {
} elsif ($self->{'block_pos'}+$rv != $self->{'block_length'}) {
# incomplete write
# Remove the written part
$self->{'stdin_buffer_pos'} += $rv;
$self->{'block_pos'} += $rv;
$something_written = $rv;
} else {
# successfully wrote everything
my $a = "";
$self->set_stdin_buffer(\$a,\$a,0,"","");
$self->set_block(\$a,\$a,0,"","");
$something_written = $rv;
}
}
@ -8001,11 +8010,13 @@ sub push {
if($perlexpr =~ /^(\d+) /) {
# Positional
defined($record->[$1-1]) or next;
$self->{'len'}{$perlexpr} += length $record->[$1-1]->replace($perlexpr,$quote_arg,$self);
$self->{'len'}{$perlexpr} +=
length $record->[$1-1]->replace($perlexpr,$quote_arg,$self);
} else {
for my $arg (@$record) {
if(defined $arg) {
$self->{'len'}{$perlexpr} += length $arg->replace($perlexpr,$quote_arg,$self);
$self->{'len'}{$perlexpr} +=
length $arg->replace($perlexpr,$quote_arg,$self);
}
}
}
@ -8604,7 +8615,7 @@ sub get {
if($opt::pipe or $opt::pipepart) {
if($cmd_line->replaced() eq "") {
# Empty command - pipe requires a command
::error("--pipe must have a command to pipe into (e.g. 'cat').");
::error("--pipe/--pipepart must have a command to pipe into (e.g. 'cat').");
::wait_and_exit(255);
}
} else {

View file

@ -426,7 +426,7 @@ string that is not in the command line.
See also: B<:::>.
=item B<--bar> (alpha testing)
=item B<--bar> (beta testing)
Show progress as a progress bar. In the bar is shown: % of jobs
completed, estimated seconds left, and number of jobs started.
@ -513,7 +513,7 @@ I<size> defaults to 1M.
See B<--pipe> and B<--pipepart> for use of this.
=item B<--cat> (alpha testing)
=item B<--cat> (beta testing)
Create a temporary file with content. Normally B<--pipe>/B<--pipepart>
will give data to the program on stdin (standard input). With B<--cat>
@ -613,7 +613,7 @@ I<secs> seconds after starting each job. I<secs> can be less than 1
second.
=item B<--dry-run> (alpha testing)
=item B<--dry-run> (beta testing)
Print the job to run on stdout (standard output), but do not run the
job. Use B<-v -v> to include the wrapping that GNU Parallel generates
@ -672,7 +672,7 @@ and functions) use env_parallel as described under the option I<command>.
See also: B<--record-env>.
=item B<--eta> (alpha testing)
=item B<--eta> (beta testing)
Show the estimated number of seconds before finishing. This forces GNU
B<parallel> to read all jobs before starting to find the number of
@ -693,7 +693,7 @@ Implies B<--semaphore>.
See also B<--bg>, B<man sem>.
=item B<--fifo> (alpha testing)
=item B<--fifo> (beta testing)
Create a temporary fifo with content. Normally B<--pipe> and
B<--pipepart> will give data to the program on stdin (standard
@ -725,7 +725,7 @@ over B<--tollef>. The B<--tollef> option is now retired, and therefore
may not be used. B<--gnu> is kept for compatibility.
=item B<--group> (alpha testing)
=item B<--group> (beta testing)
Group output. Output from each jobs is grouped together and is only
printed when the command is finished. stderr (standard error) first
@ -747,9 +747,9 @@ See also: B<--line-buffer> B<--ungroup>
Print a summary of the options to GNU B<parallel> and exit.
=item B<--halt-on-error> I<val> (alpha testing)
=item B<--halt-on-error> I<val> (beta testing)
=item B<--halt> I<val> (alpha testing)
=item B<--halt> I<val> (beta testing)
When should GNU B<parallel> terminate? In some situations it makes no
sense to run all jobs. GNU B<parallel> should simply give up as soon
@ -1172,7 +1172,7 @@ performance is important use B<--pipepart>.
See also: B<--recstart>, B<--recend>, B<--fifo>, B<--cat>, B<--pipepart>.
=item B<--pipepart> (alpha testing)
=item B<--pipepart> (beta testing)
Pipe parts of a physical file. B<--pipepart> works similar to
B<--pipe>, but is much faster. It has a few limitations:
@ -1198,7 +1198,7 @@ control on the command line (used by GNU B<parallel> internally when
called with B<--sshlogin>).
=item B<--plus> (alpha testing)
=item B<--plus> (beta testing)
Activate additional replacement strings: {+/} {+.} {+..} {+...} {..}
{...} {/..} {/...} {##}. The idea being that '{+foo}' matches the opposite of
@ -1288,7 +1288,7 @@ Overrides an earlier B<--keep-order> (e.g. if set in
B<~/.parallel/config>).
=item B<--nice> I<niceness> (alpha testing)
=item B<--nice> I<niceness> (beta testing)
Run the command at this niceness. For simple commands you can just add
B<nice> in front of the command. But if the command consists of more
@ -1737,7 +1737,7 @@ Does not run the command but quotes it. Useful for making quoted
composed commands for GNU B<parallel>.
=item B<--shuf> (alpha testing)
=item B<--shuf> (beta testing)
Shuffle jobs. When having multiple input sources it is hard to
randomize jobs. --shuf will generate all jobs, and shuffle them before
@ -1765,13 +1765,13 @@ I<secs> seconds after starting each ssh. I<secs> can be less than 1
seconds.
=item B<-S> I<[@hostgroups/][ncpu/]sshlogin[,[@hostgroups/][ncpu/]sshlogin[,...]]> (alpha testing)
=item B<-S> I<[@hostgroups/][ncpu/]sshlogin[,[@hostgroups/][ncpu/]sshlogin[,...]]> (beta testing)
=item B<-S> I<@hostgroup> (alpha testing)
=item B<-S> I<@hostgroup> (beta testing)
=item B<--sshlogin> I<[@hostgroups/][ncpu/]sshlogin[,[@hostgroups/][ncpu/]sshlogin[,...]]> (alpha testing)
=item B<--sshlogin> I<[@hostgroups/][ncpu/]sshlogin[,[@hostgroups/][ncpu/]sshlogin[,...]]> (beta testing)
=item B<--sshlogin> I<@hostgroup> (alpha testing)
=item B<--sshlogin> I<@hostgroup> (beta testing)
Distribute jobs to remote computers. The jobs will be run on a list of
remote computers.
@ -2088,9 +2088,9 @@ Use B<-v> B<-v> to print the wrapping ssh command when running remotely.
Print the version GNU B<parallel> and exit.
=item B<--workdir> I<mydir> (alpha testing)
=item B<--workdir> I<mydir> (beta testing)
=item B<--wd> I<mydir> (alpha testing)
=item B<--wd> I<mydir> (beta testing)
Files transferred using B<--transfer> and B<--return> will be relative
to I<mydir> on remote computers, and the command will be executed in

View file

@ -3,7 +3,7 @@
# SSH only allowed to localhost/lo
cat <<'EOF' | sed -e s/\$SERVER1/$SERVER1/\;s/\$SERVER2/$SERVER2/ | parallel -vj2 --retries 3 -k --joblog /tmp/jl-`basename $0` -L1
echo '### --hostgroup force ncpu'
parallel --delay 0.1 --hgrp -S @g1/1/parallel@lo -S @g2/3/lo whoami\;sleep 0.2{} ::: {1..8} | sort
parallel --delay 0.1 --hgrp -S @g1/1/parallel@lo -S @g2/3/lo whoami\;sleep 0.3{} ::: {1..8} | sort
echo '### --hostgroup two group arg'
parallel -k --sshdelay 0.1 --hgrp -S @g1/1/parallel@lo -S @g2/3/lo whoami\;sleep 0.3{} ::: {1..8}@g1+g2 | sort
@ -12,7 +12,7 @@ echo '### --hostgroup one group arg'
parallel --delay 0.1 --hgrp -S @g1/1/parallel@lo -S @g2/3/lo whoami\;sleep 0.2{} ::: {1..8}@g2
echo '### --hostgroup multiple group arg + unused group'
parallel --delay 0.1 --hgrp -S @g1/1/parallel@lo -S @g1/3/lo -S @g3/100/tcsh@lo whoami\;sleep 0.2{} ::: {1..8}@g1+g2 | sort
parallel --delay 0.1 --hgrp -S @g1/1/parallel@lo -S @g1/3/lo -S @g3/100/tcsh@lo whoami\;sleep 0.4{} ::: {1..8}@g1+g2 | sort
echo '### --hostgroup two groups @'
parallel -k --hgrp -S @g1/parallel@lo -S @g2/lo --tag whoami\;echo ::: parallel@g1 tange@g2

View file

@ -1,6 +1,6 @@
echo '### --hostgroup force ncpu'
### --hostgroup force ncpu
parallel --delay 0.1 --hgrp -S @g1/1/parallel@lo -S @g2/3/lo whoami\;sleep 0.2{} ::: {1..8} | sort
parallel --delay 0.1 --hgrp -S @g1/1/parallel@lo -S @g2/3/lo whoami\;sleep 0.3{} ::: {1..8} | sort
parallel
parallel
tange
@ -33,7 +33,7 @@ tange
tange
echo '### --hostgroup multiple group arg + unused group'
### --hostgroup multiple group arg + unused group
parallel --delay 0.1 --hgrp -S @g1/1/parallel@lo -S @g1/3/lo -S @g3/100/tcsh@lo whoami\;sleep 0.2{} ::: {1..8}@g1+g2 | sort
parallel --delay 0.1 --hgrp -S @g1/1/parallel@lo -S @g1/3/lo -S @g3/100/tcsh@lo whoami\;sleep 0.4{} ::: {1..8}@g1+g2 | sort
parallel
parallel
tange