2 # -*- mode: perl; perl-indent-level: 2; indent-tabs-mode: nil; -*-
6 crunch-job: Execute job steps, save snapshots as requested, collate output.
10 Obtain job details from Arvados, run tasks on compute nodes (typically
11 invoked by scheduler on controller):
13 crunch-job --job x-y-z --git-dir /path/to/repo/.git
15 Obtain job details from command line, run tasks on local machine
16 (typically invoked by application or developer on VM):
18 crunch-job --job '{"script_version":"/path/to/working/tree","script":"scriptname",...}'
20 crunch-job --job '{"repository":"https://github.com/curoverse/arvados.git","script_version":"master","script":"scriptname",...}'
28 If the job is already locked, steal the lock and run it anyway.
32 Path to a .git directory (or a git URL) where the commit given in the
33 job's C<script_version> attribute is to be found. If this is I<not>
34 given, the job's C<repository> attribute will be used.
38 Arvados API authorization token to use during the course of the job.
42 Do not clear per-job/task temporary directories during initial job
43 setup. This can speed up development and debugging when running jobs
48 UUID of the job to run, or a JSON-encoded job resource without a
49 UUID. If the latter is given, a new job object will be created.
53 =head1 RUNNING JOBS LOCALLY
55 crunch-job's log messages appear on stderr along with the job tasks'
56 stderr streams. The log is saved in Keep at each checkpoint and when
59 If the job succeeds, the job's output locator is printed on stdout.
61 While the job is running, the following signals are accepted:
65 =item control-C, SIGINT, SIGQUIT
67 Save a checkpoint, terminate any job tasks that are running, and stop.
71 Save a checkpoint and continue.
75 Refresh node allocation (i.e., check whether any nodes have been added
76 or unallocated) and attributes of the Job record that should affect
77 behavior (e.g., cancel job if cancelled_at becomes non-nil).
85 use POSIX ':sys_wait_h';
86 use POSIX qw(strftime);
87 use Fcntl qw(F_GETFL F_SETFL O_NONBLOCK);
91 use Digest::MD5 qw(md5_hex);
97 use File::Path qw( make_path remove_tree );
99 use constant TASK_TEMPFAIL => 111;
100 use constant EX_TEMPFAIL => 75;
101 use constant EX_RETRY_UNLOCKED => 93;
103 $ENV{"TMPDIR"} ||= "/tmp";
104 unless (defined $ENV{"CRUNCH_TMP"}) {
105 $ENV{"CRUNCH_TMP"} = $ENV{"TMPDIR"} . "/crunch-job";
106 if ($ENV{"USER"} ne "crunch" && $< != 0) {
107 # use a tmp dir unique for my uid
108 $ENV{"CRUNCH_TMP"} .= "-$<";
112 # Create the tmp directory if it does not exist
113 if ( ! -d $ENV{"CRUNCH_TMP"} ) {
114 make_path $ENV{"CRUNCH_TMP"} or die "Failed to create temporary working directory: " . $ENV{"CRUNCH_TMP"};
117 $ENV{"JOB_WORK"} = $ENV{"CRUNCH_TMP"} . "/work";
118 $ENV{"CRUNCH_INSTALL"} = "$ENV{CRUNCH_TMP}/opt";
119 $ENV{"CRUNCH_WORK"} = $ENV{"JOB_WORK"}; # deprecated
120 mkdir ($ENV{"JOB_WORK"});
129 my $docker_bin = "docker.io";
130 my $docker_run_args = "";
131 GetOptions('force-unlock' => \$force_unlock,
132 'git-dir=s' => \$git_dir,
133 'job=s' => \$jobspec,
134 'job-api-token=s' => \$job_api_token,
135 'no-clear-tmp' => \$no_clear_tmp,
136 'resume-stash=s' => \$resume_stash,
137 'docker-bin=s' => \$docker_bin,
138 'docker-run-args=s' => \$docker_run_args,
141 if (defined $job_api_token) {
142 $ENV{ARVADOS_API_TOKEN} = $job_api_token;
145 my $have_slurm = exists $ENV{SLURM_JOB_ID} && exists $ENV{SLURM_NODELIST};
150 $main::ENV{CRUNCH_DEBUG} = 1;
154 $main::ENV{CRUNCH_DEBUG} = 0;
157 my $arv = Arvados->new('apiVersion' => 'v1');
166 if ($jobspec =~ /^[-a-z\d]+$/)
168 # $jobspec is an Arvados UUID, not a JSON job specification
169 $Job = api_call("jobs/get", uuid => $jobspec);
174 $local_job = JSON::decode_json($jobspec);
178 # Make sure our workers (our slurm nodes, localhost, or whatever) are
179 # at least able to run basic commands: they aren't down or severely
182 if (($Job || $local_job)->{docker_image_locator}) {
183 $cmd = [$docker_bin, 'ps', '-q'];
185 Log(undef, "Sanity check is `@$cmd`");
186 srun(["srun", "--nodes=\Q$ENV{SLURM_NNODES}\E", "--ntasks-per-node=1"],
190 Log(undef, "Sanity check failed: ".exit_status_s($?));
193 Log(undef, "Sanity check OK");
196 my $User = api_call("users/current");
199 if (!$force_unlock) {
200 # Claim this job, and make sure nobody else does
201 eval { api_call("jobs/lock", uuid => $Job->{uuid}); };
203 Log(undef, "Error while locking job, exiting ".EX_TEMPFAIL);
212 map { croak ("No $_ specified") unless $local_job->{$_} }
213 qw(script script_version script_parameters);
216 $local_job->{'is_locked_by_uuid'} = $User->{'uuid'};
217 $local_job->{'started_at'} = gmtime;
218 $local_job->{'state'} = 'Running';
220 $Job = api_call("jobs/create", job => $local_job);
222 $job_id = $Job->{'uuid'};
224 my $keep_logfile = $job_id . '.log.txt';
225 log_writer_start($keep_logfile);
227 $Job->{'runtime_constraints'} ||= {};
228 $Job->{'runtime_constraints'}->{'max_tasks_per_node'} ||= 0;
229 my $max_ncpus = $Job->{'runtime_constraints'}->{'max_tasks_per_node'};
231 my $gem_versions = `gem list --quiet arvados-cli 2>/dev/null`;
233 $gem_versions =~ s/^arvados-cli \(/ with arvados-cli Gem version(s) /;
234 chomp($gem_versions);
235 chop($gem_versions); # Closing parentheses
240 "running from " . ((-e $0) ? realpath($0) : "stdin") . $gem_versions);
242 Log (undef, "check slurm allocation");
245 # Should use $ENV{SLURM_TASKS_PER_NODE} instead of sinfo? (eg. "4(x3),2,4(x2)")
249 my $localcpus = 0 + `grep -cw ^processor /proc/cpuinfo` || 1;
250 push @sinfo, "$localcpus localhost";
252 if (exists $ENV{SLURM_NODELIST})
254 push @sinfo, `sinfo -h --format='%c %N' --nodes=\Q$ENV{SLURM_NODELIST}\E`;
258 my ($ncpus, $slurm_nodelist) = split;
259 $ncpus = $max_ncpus if $max_ncpus && $ncpus > $max_ncpus;
262 while ($slurm_nodelist =~ s/^([^\[,]+?(\[.*?\])?)(,|$)//)
265 if ($nodelist =~ /\[((\d+)(-(\d+))?(,(\d+)(-(\d+))?)*)\]/)
268 foreach (split (",", $ranges))
281 push @nodelist, map {
283 $n =~ s/\[[-,\d]+\]/$_/;
290 push @nodelist, $nodelist;
293 foreach my $nodename (@nodelist)
295 Log (undef, "node $nodename - $ncpus slots");
296 my $node = { name => $nodename,
298 # The number of consecutive times a task has been dispatched
299 # to this node and failed.
301 # The number of consecutive times that SLURM has reported
302 # a node failure since the last successful task.
304 # Don't dispatch work to this node until this time
305 # (in seconds since the epoch) has passed.
307 foreach my $cpu (1..$ncpus)
309 push @slot, { node => $node,
313 push @node, @nodelist;
318 # Ensure that we get one jobstep running on each allocated node before
319 # we start overloading nodes with concurrent steps
321 @slot = sort { $a->{cpu} <=> $b->{cpu} } @slot;
324 $Job->update_attributes(
325 'tasks_summary' => { 'failed' => 0,
330 Log (undef, "start");
331 $SIG{'INT'} = sub { $main::please_freeze = 1; };
332 $SIG{'QUIT'} = sub { $main::please_freeze = 1; };
333 $SIG{'TERM'} = \&croak;
334 $SIG{'TSTP'} = sub { $main::please_freeze = 1; };
335 $SIG{'ALRM'} = sub { $main::please_info = 1; };
336 $SIG{'CONT'} = sub { $main::please_continue = 1; };
337 $SIG{'HUP'} = sub { $main::please_refresh = 1; };
339 $main::please_freeze = 0;
340 $main::please_info = 0;
341 $main::please_continue = 0;
342 $main::please_refresh = 0;
343 my $jobsteps_must_output_keys = 0; # becomes 1 when any task outputs a key
345 grep { $ENV{$1} = $2 if /^(NOCACHE.*?)=(.*)/ } split ("\n", $$Job{knobs});
346 $ENV{"CRUNCH_JOB_UUID"} = $job_id;
347 $ENV{"JOB_UUID"} = $job_id;
350 my @jobstep_todo = ();
351 my @jobstep_done = ();
352 my @jobstep_tomerge = ();
353 my $jobstep_tomerge_level = 0;
354 my $squeue_checked = 0;
355 my $latest_refresh = scalar time;
359 if (defined $Job->{thawedfromkey})
361 thaw ($Job->{thawedfromkey});
365 my $first_task = api_call("job_tasks/create", job_task => {
366 'job_uuid' => $Job->{'uuid'},
371 push @jobstep, { 'level' => 0,
373 'arvados_task' => $first_task,
375 push @jobstep_todo, 0;
381 must_lock_now("$ENV{CRUNCH_TMP}/.lock", "a job is already running here.");
384 my $build_script = handle_readall(\*DATA);
385 my $nodelist = join(",", @node);
386 my $git_tar_count = 0;
388 if (!defined $no_clear_tmp) {
389 # Clean out crunch_tmp/work, crunch_tmp/opt, crunch_tmp/src*
390 Log (undef, "Clean work dirs");
392 my $cleanpid = fork();
395 # Find FUSE mounts under $CRUNCH_TMP and unmount them.
396 # Then clean up work directories.
397 # TODO: When #5036 is done and widely deployed, we can limit mount's
398 # -t option to simply fuse.keep.
399 srun (["srun", "--nodelist=$nodelist", "-D", $ENV{'TMPDIR'}],
400 ['bash', '-ec', '-o', 'pipefail', 'mount -t fuse,fuse.keep | awk "(index(\$3, \"$CRUNCH_TMP\") == 1){print \$3}" | xargs -r -n 1 fusermount -u -z; sleep 1; rm -rf $JOB_WORK $CRUNCH_INSTALL $CRUNCH_TMP/task $CRUNCH_TMP/src* $CRUNCH_TMP/*.cid']);
405 last if $cleanpid == waitpid (-1, WNOHANG);
406 freeze_if_want_freeze ($cleanpid);
407 select (undef, undef, undef, 0.1);
410 Log(undef, "Clean work dirs: exit ".exit_status_s($?));
411 exit(EX_RETRY_UNLOCKED);
415 # If this job requires a Docker image, install that.
416 my ($docker_locator, $docker_stream, $docker_hash, $docker_limitmem, $dockeruserarg);
417 if ($docker_locator = $Job->{docker_image_locator}) {
418 ($docker_stream, $docker_hash) = find_docker_image($docker_locator);
421 croak("No Docker image hash found from locator $docker_locator");
423 $docker_stream =~ s/^\.//;
424 my $docker_install_script = qq{
425 if ! $docker_bin images -q --no-trunc --all | grep -qxF \Q$docker_hash\E; then
426 arv-get \Q$docker_locator$docker_stream/$docker_hash.tar\E | $docker_bin load
429 my $docker_pid = fork();
430 if ($docker_pid == 0)
432 srun (["srun", "--nodelist=" . join(',', @node)],
433 ["/bin/sh", "-ec", $docker_install_script]);
438 last if $docker_pid == waitpid (-1, WNOHANG);
439 freeze_if_want_freeze ($docker_pid);
440 select (undef, undef, undef, 0.1);
444 croak("Installing Docker image from $docker_locator exited "
448 # Determine whether this version of Docker supports memory+swap limits.
449 srun(["srun", "--nodelist=" . $node[0]],
450 ["/bin/sh", "-ec", "$docker_bin run --help | grep -qe --memory-swap="],
452 $docker_limitmem = ($? == 0);
454 # Find a non-root Docker user to use.
455 # Tries the default user for the container, then 'crunch', then 'nobody',
456 # testing for whether the actual user id is non-zero. This defends against
457 # mistakes but not malice, but we intend to harden the security in the future
458 # so we don't want anyone getting used to their jobs running as root in their
460 my @tryusers = ("", "crunch", "nobody");
461 foreach my $try_user (@tryusers) {
463 if ($try_user eq "") {
464 Log(undef, "Checking if container default user is not UID 0");
467 Log(undef, "Checking if user '$try_user' is not UID 0");
468 $try_user_arg = "--user=$try_user";
470 srun(["srun", "--nodelist=" . $node[0]],
472 "a=`$docker_bin run $docker_run_args $try_user_arg $docker_hash id --user` && " .
476 $dockeruserarg = $try_user_arg;
477 if ($try_user eq "") {
478 Log(undef, "Container will run with default user");
480 Log(undef, "Container will run with $dockeruserarg");
486 if (!defined $dockeruserarg) {
487 croak("Could not find a user in container that is not UID 0 (tried default user, @tryusers) or there was a problem running 'id' in the container.");
490 if ($Job->{arvados_sdk_version}) {
491 # The job also specifies an Arvados SDK version. Add the SDKs to the
492 # tar file for the build script to install.
493 Log(undef, sprintf("Packing Arvados SDK version %s for installation",
494 $Job->{arvados_sdk_version}));
495 add_git_archive("git", "--git-dir=$git_dir", "archive",
496 "--prefix=.arvados.sdk/",
497 $Job->{arvados_sdk_version}, "sdk");
501 if (!defined $git_dir && $Job->{'script_version'} =~ m{^/}) {
502 # If script_version looks like an absolute path, *and* the --git-dir
503 # argument was not given -- which implies we were not invoked by
504 # crunch-dispatch -- we will use the given path as a working
505 # directory instead of resolving script_version to a git commit (or
506 # doing anything else with git).
507 $ENV{"CRUNCH_SRC_COMMIT"} = $Job->{'script_version'};
508 $ENV{"CRUNCH_SRC"} = $Job->{'script_version'};
511 # Resolve the given script_version to a git commit sha1. Also, if
512 # the repository is remote, clone it into our local filesystem: this
513 # ensures "git archive" will work, and is necessary to reliably
514 # resolve a symbolic script_version like "master^".
515 $ENV{"CRUNCH_SRC"} = "$ENV{CRUNCH_TMP}/src";
517 Log (undef, "Looking for version ".$Job->{script_version}." from repository ".$Job->{repository});
519 $ENV{"CRUNCH_SRC_COMMIT"} = $Job->{script_version};
521 # If we're running under crunch-dispatch, it will have already
522 # pulled the appropriate source tree into its own repository, and
523 # given us that repo's path as $git_dir.
525 # If we're running a "local" job, we might have to fetch content
526 # from a remote repository.
528 # (Currently crunch-dispatch gives a local path with --git-dir, but
529 # we might as well accept URLs there too in case it changes its
531 my $repo = $git_dir || $Job->{'repository'};
533 # Repository can be remote or local. If remote, we'll need to fetch it
534 # to a local dir before doing `git log` et al.
537 if ($repo =~ m{://|^[^/]*:}) {
538 # $repo is a git url we can clone, like git:// or https:// or
539 # file:/// or [user@]host:repo.git. Note "user/name@host:foo" is
540 # not recognized here because distinguishing that from a local
541 # path is too fragile. If you really need something strange here,
542 # use the ssh:// form.
543 $repo_location = 'remote';
544 } elsif ($repo =~ m{^\.*/}) {
545 # $repo is a local path to a git index. We'll also resolve ../foo
546 # to ../foo/.git if the latter is a directory. To help
547 # disambiguate local paths from named hosted repositories, this
548 # form must be given as ./ or ../ if it's a relative path.
549 if (-d "$repo/.git") {
550 $repo = "$repo/.git";
552 $repo_location = 'local';
554 # $repo is none of the above. It must be the name of a hosted
556 my $arv_repo_list = api_call("repositories/list",
557 'filters' => [['name','=',$repo]]);
558 my @repos_found = @{$arv_repo_list->{'items'}};
559 my $n_found = $arv_repo_list->{'serverResponse'}->{'items_available'};
561 Log(undef, "Repository '$repo' -> "
562 . join(", ", map { $_->{'uuid'} } @repos_found));
565 croak("Error: Found $n_found repositories with name '$repo'.");
567 $repo = $repos_found[0]->{'fetch_url'};
568 $repo_location = 'remote';
570 Log(undef, "Using $repo_location repository '$repo'");
571 $ENV{"CRUNCH_SRC_URL"} = $repo;
573 # Resolve given script_version (we'll call that $treeish here) to a
574 # commit sha1 ($commit).
575 my $treeish = $Job->{'script_version'};
577 if ($repo_location eq 'remote') {
578 # We minimize excess object-fetching by re-using the same bare
579 # repository in CRUNCH_TMP/.git for multiple crunch-jobs -- we
580 # just keep adding remotes to it as needed.
581 my $local_repo = $ENV{'CRUNCH_TMP'}."/.git";
582 my $gitcmd = "git --git-dir=\Q$local_repo\E";
584 # Set up our local repo for caching remote objects, making
586 if (!-d $local_repo) {
587 make_path($local_repo) or croak("Error: could not create $local_repo");
589 # This works (exits 0 and doesn't delete fetched objects) even
590 # if $local_repo is already initialized:
591 `$gitcmd init --bare`;
593 croak("Error: $gitcmd init --bare exited ".exit_status_s($?));
596 # If $treeish looks like a hash (or abbrev hash) we look it up in
597 # our local cache first, since that's cheaper. (We don't want to
598 # do that with tags/branches though -- those change over time, so
599 # they should always be resolved by the remote repo.)
600 if ($treeish =~ /^[0-9a-f]{7,40}$/s) {
601 # Hide stderr because it's normal for this to fail:
602 my $sha1 = `$gitcmd rev-list -n1 ''\Q$treeish\E 2>/dev/null`;
604 # Careful not to resolve a branch named abcdeff to commit 1234567:
605 $sha1 =~ /^$treeish/ &&
606 $sha1 =~ /^([0-9a-f]{40})$/s) {
608 Log(undef, "Commit $commit already present in $local_repo");
612 if (!defined $commit) {
613 # If $treeish isn't just a hash or abbrev hash, or isn't here
614 # yet, we need to fetch the remote to resolve it correctly.
616 # First, remove all local heads. This prevents a name that does
617 # not exist on the remote from resolving to (or colliding with)
618 # a previously fetched branch or tag (possibly from a different
620 remove_tree("$local_repo/refs/heads", {keep_root => 1});
622 Log(undef, "Fetching objects from $repo to $local_repo");
623 `$gitcmd fetch --no-progress --tags ''\Q$repo\E \Q+refs/heads/*:refs/heads/*\E`;
625 croak("Error: `$gitcmd fetch` exited ".exit_status_s($?));
629 # Now that the data is all here, we will use our local repo for
630 # the rest of our git activities.
634 my $gitcmd = "git --git-dir=\Q$repo\E";
635 my $sha1 = `$gitcmd rev-list -n1 ''\Q$treeish\E`;
636 unless ($? == 0 && $sha1 =~ /^([0-9a-f]{40})$/) {
637 croak("`$gitcmd rev-list` exited "
639 .", '$treeish' not found, giving up");
642 Log(undef, "Version $treeish is commit $commit");
644 if ($commit ne $Job->{'script_version'}) {
645 # Record the real commit id in the database, frozentokey, logs,
646 # etc. -- instead of an abbreviation or a branch name which can
647 # become ambiguous or point to a different commit in the future.
648 if (!$Job->update_attributes('script_version' => $commit)) {
649 croak("Error: failed to update job's script_version attribute");
653 $ENV{"CRUNCH_SRC_COMMIT"} = $commit;
654 add_git_archive("$gitcmd archive ''\Q$commit\E");
657 my $git_archive = combined_git_archive();
658 if (!defined $git_archive) {
659 Log(undef, "Skip install phase (no git archive)");
661 Log(undef, "Warning: This probably means workers have no source tree!");
666 my $install_script_tries_left = 3;
667 for (my $attempts = 0; $attempts < 3; $attempts++) {
668 Log(undef, "Run install script on all workers");
670 my @srunargs = ("srun",
671 "--nodelist=$nodelist",
672 "-D", $ENV{'TMPDIR'}, "--job-name=$job_id");
673 my @execargs = ("sh", "-c",
674 "mkdir -p $ENV{CRUNCH_INSTALL} && cd $ENV{CRUNCH_TMP} && perl -");
676 $ENV{"CRUNCH_GIT_ARCHIVE_HASH"} = md5_hex($git_archive);
677 my ($install_stderr_r, $install_stderr_w);
678 pipe $install_stderr_r, $install_stderr_w or croak("pipe() failed: $!");
679 set_nonblocking($install_stderr_r);
680 my $installpid = fork();
681 if ($installpid == 0)
683 close($install_stderr_r);
684 fcntl($install_stderr_w, F_SETFL, 0) or croak($!); # no close-on-exec
685 open(STDOUT, ">&", $install_stderr_w);
686 open(STDERR, ">&", $install_stderr_w);
687 srun (\@srunargs, \@execargs, {}, $build_script . $git_archive);
690 close($install_stderr_w);
691 # Tell freeze_if_want_freeze how to kill the child, otherwise the
692 # "waitpid(installpid)" loop won't get interrupted by a freeze:
693 $proc{$installpid} = {};
695 # Track whether anything appears on stderr other than slurm errors
696 # ("srun: ...") and the "starting: ..." message printed by the
697 # srun subroutine itself:
698 my $stderr_anything_from_script = 0;
699 my $match_our_own_errors = '^(srun: error: |starting: \[)';
700 while ($installpid != waitpid(-1, WNOHANG)) {
701 freeze_if_want_freeze ($installpid);
702 # Wait up to 0.1 seconds for something to appear on stderr, then
703 # do a non-blocking read.
704 my $bits = fhbits($install_stderr_r);
705 select ($bits, undef, $bits, 0.1);
706 if (0 < sysread ($install_stderr_r, $stderr_buf, 8192, length($stderr_buf)))
708 while ($stderr_buf =~ /^(.*?)\n/) {
710 substr $stderr_buf, 0, 1+length($line), "";
711 Log(undef, "stderr $line");
712 if ($line !~ /$match_our_own_errors/) {
713 $stderr_anything_from_script = 1;
718 delete $proc{$installpid};
719 $install_exited = $?;
720 close($install_stderr_r);
721 if (length($stderr_buf) > 0) {
722 if ($stderr_buf !~ /$match_our_own_errors/) {
723 $stderr_anything_from_script = 1;
725 Log(undef, "stderr $stderr_buf")
728 Log (undef, "Install script exited ".exit_status_s($install_exited));
729 last if $install_exited == 0 || $main::please_freeze;
730 # If the install script fails but doesn't print an error message,
731 # the next thing anyone is likely to do is just run it again in
732 # case it was a transient problem like "slurm communication fails
733 # because the network isn't reliable enough". So we'll just do
734 # that ourselves (up to 3 attempts in total). OTOH, if there is an
735 # error message, the problem is more likely to have a real fix and
736 # we should fail the job so the fixing process can start, instead
737 # of doing 2 more attempts.
738 last if $stderr_anything_from_script;
741 foreach my $tar_filename (map { tar_filename_n($_); } (1..$git_tar_count)) {
742 unlink($tar_filename);
745 if ($install_exited != 0) {
750 foreach (qw (script script_version script_parameters runtime_constraints))
754 (ref($Job->{$_}) ? JSON::encode_json($Job->{$_}) : $Job->{$_}));
756 foreach (split (/\n/, $Job->{knobs}))
758 Log (undef, "knob " . $_);
763 $main::success = undef;
769 my $thisround_succeeded = 0;
770 my $thisround_failed = 0;
771 my $thisround_failed_multiple = 0;
772 my $working_slot_count = scalar(@slot);
774 @jobstep_todo = sort { $jobstep[$a]->{level} <=> $jobstep[$b]->{level}
775 or $a <=> $b } @jobstep_todo;
776 my $level = $jobstep[$jobstep_todo[0]]->{level};
778 my $initial_tasks_this_level = 0;
779 foreach my $id (@jobstep_todo) {
780 $initial_tasks_this_level++ if ($jobstep[$id]->{level} == $level);
783 # If the number of tasks scheduled at this level #T is smaller than the number
784 # of slots available #S, only use the first #T slots, or the first slot on
785 # each node, whichever number is greater.
787 # When we dispatch tasks later, we'll allocate whole-node resources like RAM
788 # based on these numbers. Using fewer slots makes more resources available
789 # to each individual task, which should normally be a better strategy when
790 # there are fewer of them running with less parallelism.
792 # Note that this calculation is not redone if the initial tasks at
793 # this level queue more tasks at the same level. This may harm
794 # overall task throughput for that level.
796 if ($initial_tasks_this_level < @node) {
797 @freeslot = (0..$#node);
798 } elsif ($initial_tasks_this_level < @slot) {
799 @freeslot = (0..$initial_tasks_this_level - 1);
801 @freeslot = (0..$#slot);
803 my $round_num_freeslots = scalar(@freeslot);
805 my %round_max_slots = ();
806 for (my $ii = $#freeslot; $ii >= 0; $ii--) {
807 my $this_slot = $slot[$freeslot[$ii]];
808 my $node_name = $this_slot->{node}->{name};
809 $round_max_slots{$node_name} ||= $this_slot->{cpu};
810 last if (scalar(keys(%round_max_slots)) >= @node);
813 Log(undef, "start level $level with $round_num_freeslots slots");
816 my $progress_is_dirty = 1;
817 my $progress_stats_updated = 0;
819 update_progress_stats();
823 for (my $todo_ptr = 0; $todo_ptr <= $#jobstep_todo; $todo_ptr ++)
825 # Don't create new tasks if we already know the job's final result.
826 last if defined($main::success);
828 my $id = $jobstep_todo[$todo_ptr];
829 my $Jobstep = $jobstep[$id];
830 if ($Jobstep->{level} != $level)
835 pipe $reader{$id}, "writer" or croak("pipe() failed: $!");
836 set_nonblocking($reader{$id});
838 my $childslot = $freeslot[0];
839 my $childnode = $slot[$childslot]->{node};
840 my $childslotname = join (".",
841 $slot[$childslot]->{node}->{name},
842 $slot[$childslot]->{cpu});
844 my $childpid = fork();
847 $SIG{'INT'} = 'DEFAULT';
848 $SIG{'QUIT'} = 'DEFAULT';
849 $SIG{'TERM'} = 'DEFAULT';
851 foreach (values (%reader))
855 fcntl ("writer", F_SETFL, 0) or croak ($!); # no close-on-exec
856 open(STDOUT,">&writer");
857 open(STDERR,">&writer");
862 delete $ENV{"GNUPGHOME"};
863 $ENV{"TASK_UUID"} = $Jobstep->{'arvados_task'}->{'uuid'};
864 $ENV{"TASK_QSEQUENCE"} = $id;
865 $ENV{"TASK_SEQUENCE"} = $level;
866 $ENV{"JOB_SCRIPT"} = $Job->{script};
867 while (my ($param, $value) = each %{$Job->{script_parameters}}) {
868 $param =~ tr/a-z/A-Z/;
869 $ENV{"JOB_PARAMETER_$param"} = $value;
871 $ENV{"TASK_SLOT_NODE"} = $slot[$childslot]->{node}->{name};
872 $ENV{"TASK_SLOT_NUMBER"} = $slot[$childslot]->{cpu};
873 $ENV{"TASK_WORK"} = $ENV{"CRUNCH_TMP"}."/task/$childslotname";
874 $ENV{"HOME"} = $ENV{"TASK_WORK"};
875 $ENV{"TASK_TMPDIR"} = $ENV{"TASK_WORK"}; # deprecated
876 $ENV{"CRUNCH_NODE_SLOTS"} = $round_max_slots{$ENV{TASK_SLOT_NODE}};
877 $ENV{"PATH"} = $ENV{"CRUNCH_INSTALL"} . "/bin:" . $ENV{"PATH"};
879 my $keep_mnt = $ENV{"TASK_WORK"}.".keep";
885 "--nodelist=".$childnode->{name},
886 qw(-n1 -c1 -N1 -D), $ENV{'TMPDIR'},
887 "--job-name=$job_id.$id.$$",
890 my $stdbuf = " stdbuf --output=0 --error=0 ";
892 my $arv_file_cache = "";
893 if (defined($Job->{'runtime_constraints'}->{'keep_cache_mb_per_task'})) {
894 $arv_file_cache = "--file-cache=" . ($Job->{'runtime_constraints'}->{'keep_cache_mb_per_task'} * 1024 * 1024);
898 "if [ -e \Q$ENV{TASK_WORK}\E ]; then rm -rf \Q$ENV{TASK_WORK}\E; fi; "
899 ."mkdir -p \Q$ENV{CRUNCH_TMP}\E \Q$ENV{JOB_WORK}\E \Q$ENV{TASK_WORK}\E \Q$keep_mnt\E "
900 ."&& cd \Q$ENV{CRUNCH_TMP}\E "
901 # These environment variables get used explicitly later in
902 # $command. No tool is expected to read these values directly.
903 .q{&& MEM=$(awk '($1 == "MemTotal:"){print $2}' </proc/meminfo) }
904 .q{&& SWAP=$(awk '($1 == "SwapTotal:"){print $2}' </proc/meminfo) }
905 ."&& MEMLIMIT=\$(( (\$MEM * 95) / ($ENV{CRUNCH_NODE_SLOTS} * 100) )) "
906 ."&& let SWAPLIMIT=\$MEMLIMIT+\$SWAP ";
908 $command .= "&& exec arv-mount --read-write --mount-by-pdh=by_pdh --mount-tmp=tmp --crunchstat-interval=10 --allow-other $arv_file_cache \Q$keep_mnt\E --exec ";
909 $ENV{TASK_KEEPMOUNT} = "$keep_mnt/by_pdh";
910 $ENV{TASK_KEEPMOUNT_TMP} = "$keep_mnt/tmp";
914 my $containername = "$Jobstep->{arvados_task}->{uuid}-$Jobstep->{failures}";
915 my $cidfile = "$ENV{CRUNCH_TMP}/$containername.cid";
916 $command .= "crunchstat -cgroup-root=/sys/fs/cgroup -cgroup-parent=docker -cgroup-cid=$cidfile -poll=10000 ";
917 $command .= "$docker_bin run $docker_run_args --name=$containername --attach=stdout --attach=stderr --attach=stdin -i \Q$dockeruserarg\E --cidfile=$cidfile --sig-proxy ";
918 # We only set memory limits if Docker lets us limit both memory and swap.
919 # Memory limits alone have been supported longer, but subprocesses tend
920 # to get SIGKILL if they exceed that without any swap limit set.
921 # See #5642 for additional background.
922 if ($docker_limitmem) {
923 $command .= "--memory=\${MEMLIMIT}k --memory-swap=\${SWAPLIMIT}k ";
926 # The source tree and $destdir directory (which we have
927 # installed on the worker host) are available in the container,
928 # under the same path.
929 $command .= "--volume=\Q$ENV{CRUNCH_SRC}:$ENV{CRUNCH_SRC}:ro\E ";
930 $command .= "--volume=\Q$ENV{CRUNCH_INSTALL}:$ENV{CRUNCH_INSTALL}:ro\E ";
932 # Currently, we make the "by_pdh" directory in arv-mount's mount
933 # point appear at /keep inside the container (instead of using
934 # the same path as the host like we do with CRUNCH_SRC and
935 # CRUNCH_INSTALL). However, crunch scripts and utilities must
936 # not rely on this. They must use $TASK_KEEPMOUNT.
937 $command .= "--volume=\Q$ENV{TASK_KEEPMOUNT}:/keep:ro\E ";
938 $ENV{TASK_KEEPMOUNT} = "/keep";
940 # Ditto TASK_KEEPMOUNT_TMP, as /keep_tmp.
941 $command .= "--volume=\Q$ENV{TASK_KEEPMOUNT_TMP}:/keep_tmp\E ";
942 $ENV{TASK_KEEPMOUNT_TMP} = "/keep_tmp";
944 # TASK_WORK is almost exactly like a docker data volume: it
945 # starts out empty, is writable, and persists until no
946 # containers use it any more. We don't use --volumes-from to
947 # share it with other containers: it is only accessible to this
948 # task, and it goes away when this task stops.
950 # However, a docker data volume is writable only by root unless
951 # the mount point already happens to exist in the container with
952 # different permissions. Therefore, we [1] assume /tmp already
953 # exists in the image and is writable by the crunch user; [2]
954 # avoid putting TASK_WORK inside CRUNCH_TMP (which won't be
955 # writable if they are created by docker while setting up the
956 # other --volumes); and [3] create $TASK_WORK inside the
957 # container using $build_script.
958 $command .= "--volume=/tmp ";
959 $ENV{"TASK_WORK"} = "/tmp/crunch-job-task-work/$childslotname";
960 $ENV{"HOME"} = $ENV{"TASK_WORK"};
961 $ENV{"TASK_TMPDIR"} = $ENV{"TASK_WORK"}; # deprecated
963 # TODO: Share a single JOB_WORK volume across all task
964 # containers on a given worker node, and delete it when the job
965 # ends (and, in case that doesn't work, when the next job
968 # For now, use the same approach as TASK_WORK above.
969 $ENV{"JOB_WORK"} = "/tmp/crunch-job-work";
971 while (my ($env_key, $env_val) = each %ENV)
973 if ($env_key =~ /^(ARVADOS|CRUNCH|JOB|TASK)_/) {
974 $command .= "--env=\Q$env_key=$env_val\E ";
977 $command .= "--env=\QHOME=$ENV{HOME}\E ";
978 $command .= "\Q$docker_hash\E ";
980 if ($Job->{arvados_sdk_version}) {
982 $command .= "perl - \Q$ENV{CRUNCH_SRC}/crunch_scripts/$Job->{script}\E";
984 $command .= "/bin/sh -c \'python -c " .
985 '"from pkg_resources import get_distribution as get; print \"Using Arvados SDK version\", get(\"arvados-python-client\").version"' .
986 ">&2 2>/dev/null; " .
987 "mkdir -p \"$ENV{JOB_WORK}\" \"$ENV{TASK_WORK}\" && " .
988 "if which stdbuf >/dev/null ; then " .
989 " exec $stdbuf \Q$ENV{CRUNCH_SRC}/crunch_scripts/$Job->{script}\E ;" .
991 " exec \Q$ENV{CRUNCH_SRC}/crunch_scripts/$Job->{script}\E ;" .
996 $command .= "crunchstat -cgroup-root=/sys/fs/cgroup -poll=10000 ";
998 $command .= "perl - $ENV{CRUNCH_SRC}/crunch_scripts/" . $Job->{"script"};
1001 my @execargs = ('bash', '-c', $command);
1002 srun (\@srunargs, \@execargs, undef, $build_script);
1003 # exec() failed, we assume nothing happened.
1004 die "srun() failed on build script\n";
1007 if (!defined $childpid)
1010 delete $reader{$id};
1014 $proc{$childpid} = { jobstep => $id,
1017 jobstepname => "$job_id.$id.$childpid",
1019 croak ("assert failed: \$slot[$childslot]->{'pid'} exists") if exists $slot[$childslot]->{pid};
1020 $slot[$childslot]->{pid} = $childpid;
1022 Log ($id, "job_task ".$Jobstep->{'arvados_task'}->{'uuid'});
1023 Log ($id, "child $childpid started on $childslotname");
1024 $Jobstep->{starttime} = time;
1025 $Jobstep->{node} = $childnode->{name};
1026 $Jobstep->{slotindex} = $childslot;
1027 delete $Jobstep->{stderr};
1028 delete $Jobstep->{finishtime};
1029 delete $Jobstep->{tempfail};
1031 $Jobstep->{'arvados_task'}->{started_at} = strftime "%Y-%m-%dT%H:%M:%SZ", gmtime($Jobstep->{starttime});
1032 $Jobstep->{'arvados_task'}->save;
1034 splice @jobstep_todo, $todo_ptr, 1;
1037 $progress_is_dirty = 1;
1041 ($round_num_freeslots > @freeslot && $todo_ptr+1 > $#jobstep_todo))
1043 last THISROUND if $main::please_freeze;
1044 if ($main::please_info)
1046 $main::please_info = 0;
1048 create_output_collection();
1050 update_progress_stats();
1055 if (!$gotsome || ($latest_refresh + 2 < scalar time))
1057 check_refresh_wanted();
1059 update_progress_stats();
1060 select (undef, undef, undef, 0.1);
1062 elsif (time - $progress_stats_updated >= 30 || $progress_is_dirty)
1064 update_progress_stats();
1066 $working_slot_count = scalar(grep { $_->{node}->{fail_count} == 0 &&
1067 $_->{node}->{hold_count} < 4 } @slot);
1068 if (($thisround_failed_multiple >= 8 && $thisround_succeeded == 0) ||
1069 ($thisround_failed_multiple >= 16 && $thisround_failed_multiple > $thisround_succeeded))
1071 my $message = "Repeated failure rate too high ($thisround_failed_multiple/"
1072 .($thisround_failed+$thisround_succeeded)
1073 .") -- giving up on this round";
1074 Log (undef, $message);
1078 # move slots from freeslot to holdslot (or back to freeslot) if necessary
1079 for (my $i=$#freeslot; $i>=0; $i--) {
1080 if ($slot[$freeslot[$i]]->{node}->{hold_until} > scalar time) {
1081 push @holdslot, (splice @freeslot, $i, 1);
1084 for (my $i=$#holdslot; $i>=0; $i--) {
1085 if ($slot[$holdslot[$i]]->{node}->{hold_until} <= scalar time) {
1086 push @freeslot, (splice @holdslot, $i, 1);
1090 # give up if no nodes are succeeding
1091 if ($working_slot_count < 1) {
1092 Log(undef, "Every node has failed -- giving up");
1099 push @freeslot, splice @holdslot;
1100 map { $slot[$freeslot[$_]]->{node}->{losing_streak} = 0 } (0..$#freeslot);
1103 Log (undef, "wait for last ".(scalar keys %proc)." children to finish");
1106 if ($main::please_continue) {
1107 $main::please_continue = 0;
1110 $main::please_info = 0, freeze(), create_output_collection(), save_meta(1) if $main::please_info;
1112 if (!reapchildren())
1114 check_refresh_wanted();
1116 update_progress_stats();
1117 select (undef, undef, undef, 0.1);
1118 killem (keys %proc) if $main::please_freeze;
1122 update_progress_stats();
1123 freeze_if_want_freeze();
1126 if (!defined $main::success)
1128 if (!@jobstep_todo) {
1130 } elsif ($working_slot_count < 1) {
1131 save_output_collection();
1133 exit(EX_RETRY_UNLOCKED);
1134 } elsif ($thisround_succeeded == 0 &&
1135 ($thisround_failed == 0 || $thisround_failed > 4)) {
1136 my $message = "stop because $thisround_failed tasks failed and none succeeded";
1137 Log (undef, $message);
1142 goto ONELEVEL if !defined $main::success;
1145 release_allocation();
1147 my $collated_output = save_output_collection();
1148 Log (undef, "finish");
1153 if ($collated_output && $main::success) {
1154 $final_state = 'Complete';
1156 $final_state = 'Failed';
1158 $Job->update_attributes('state' => $final_state);
1160 exit (($final_state eq 'Complete') ? 0 : 1);
1164 sub update_progress_stats
1166 $progress_stats_updated = time;
1167 return if !$progress_is_dirty;
1168 my ($todo, $done, $running) = (scalar @jobstep_todo,
1169 scalar @jobstep_done,
1170 scalar keys(%proc));
1171 $Job->{'tasks_summary'} ||= {};
1172 $Job->{'tasks_summary'}->{'todo'} = $todo;
1173 $Job->{'tasks_summary'}->{'done'} = $done;
1174 $Job->{'tasks_summary'}->{'running'} = $running;
1175 $Job->update_attributes('tasks_summary' => $Job->{'tasks_summary'});
1176 Log (undef, "status: $done done, $running running, $todo todo");
1177 $progress_is_dirty = 0;
1184 my $pid = waitpid (-1, WNOHANG);
1185 return 0 if $pid <= 0;
1187 my $whatslot = ($slot[$proc{$pid}->{slot}]->{node}->{name}
1189 . $slot[$proc{$pid}->{slot}]->{cpu});
1190 my $jobstepid = $proc{$pid}->{jobstep};
1191 my $elapsed = time - $proc{$pid}->{time};
1192 my $Jobstep = $jobstep[$jobstepid];
1194 my $childstatus = $?;
1195 my $exitvalue = $childstatus >> 8;
1196 my $exitinfo = "exit ".exit_status_s($childstatus);
1197 $Jobstep->{'arvados_task'}->reload;
1198 my $task_success = $Jobstep->{'arvados_task'}->{success};
1200 Log ($jobstepid, "child $pid on $whatslot $exitinfo success=$task_success");
1202 if (!defined $task_success) {
1203 # task did not indicate one way or the other --> fail
1204 Log($jobstepid, sprintf(
1205 "ERROR: Task process exited %s, but never updated its task record to indicate success and record its output.",
1206 exit_status_s($childstatus)));
1207 $Jobstep->{'arvados_task'}->{success} = 0;
1208 $Jobstep->{'arvados_task'}->save;
1215 $temporary_fail ||= $Jobstep->{tempfail};
1216 $temporary_fail ||= ($exitvalue == TASK_TEMPFAIL);
1218 ++$thisround_failed;
1219 ++$thisround_failed_multiple if $Jobstep->{'failures'} >= 1;
1221 # Check for signs of a failed or misconfigured node
1222 if (++$slot[$proc{$pid}->{slot}]->{node}->{losing_streak} >=
1223 2+$slot[$proc{$pid}->{slot}]->{node}->{ncpus}) {
1224 # Don't count this against jobstep failure thresholds if this
1225 # node is already suspected faulty and srun exited quickly
1226 if ($slot[$proc{$pid}->{slot}]->{node}->{hold_until} &&
1228 Log ($jobstepid, "blaming failure on suspect node " .
1229 $slot[$proc{$pid}->{slot}]->{node}->{name});
1230 $temporary_fail ||= 1;
1232 ban_node_by_slot($proc{$pid}->{slot});
1235 Log ($jobstepid, sprintf('failure (#%d, %s) after %d seconds',
1236 ++$Jobstep->{'failures'},
1237 $temporary_fail ? 'temporary' : 'permanent',
1240 if (!$temporary_fail || $Jobstep->{'failures'} >= 3) {
1241 # Give up on this task, and the whole job
1244 # Put this task back on the todo queue
1245 push @jobstep_todo, $jobstepid;
1246 $Job->{'tasks_summary'}->{'failed'}++;
1250 ++$thisround_succeeded;
1251 $slot[$proc{$pid}->{slot}]->{node}->{losing_streak} = 0;
1252 $slot[$proc{$pid}->{slot}]->{node}->{hold_until} = 0;
1253 $slot[$proc{$pid}->{slot}]->{node}->{fail_count} = 0;
1254 push @jobstep_done, $jobstepid;
1255 Log ($jobstepid, "success in $elapsed seconds");
1257 $Jobstep->{exitcode} = $childstatus;
1258 $Jobstep->{finishtime} = time;
1259 $Jobstep->{'arvados_task'}->{finished_at} = strftime "%Y-%m-%dT%H:%M:%SZ", gmtime($Jobstep->{finishtime});
1260 $Jobstep->{'arvados_task'}->save;
1261 process_stderr ($jobstepid, $task_success);
1262 Log ($jobstepid, sprintf("task output (%d bytes): %s",
1263 length($Jobstep->{'arvados_task'}->{output}),
1264 $Jobstep->{'arvados_task'}->{output}));
1266 close $reader{$jobstepid};
1267 delete $reader{$jobstepid};
1268 delete $slot[$proc{$pid}->{slot}]->{pid};
1269 push @freeslot, $proc{$pid}->{slot};
1272 if ($task_success) {
1274 my $newtask_list = [];
1275 my $newtask_results;
1277 $newtask_results = api_call(
1280 'created_by_job_task_uuid' => $Jobstep->{'arvados_task'}->{uuid}
1282 'order' => 'qsequence',
1283 'offset' => scalar(@$newtask_list),
1285 push(@$newtask_list, @{$newtask_results->{items}});
1286 } while (@{$newtask_results->{items}});
1287 foreach my $arvados_task (@$newtask_list) {
1289 'level' => $arvados_task->{'sequence'},
1291 'arvados_task' => $arvados_task
1293 push @jobstep, $jobstep;
1294 push @jobstep_todo, $#jobstep;
1298 $progress_is_dirty = 1;
1302 sub check_refresh_wanted
1304 my @stat = stat $ENV{"CRUNCH_REFRESH_TRIGGER"};
1305 if (@stat && $stat[9] > $latest_refresh) {
1306 $latest_refresh = scalar time;
1307 my $Job2 = api_call("jobs/get", uuid => $jobspec);
1308 for my $attr ('cancelled_at',
1309 'cancelled_by_user_uuid',
1310 'cancelled_by_client_uuid',
1312 $Job->{$attr} = $Job2->{$attr};
1314 if ($Job->{'state'} ne "Running") {
1315 if ($Job->{'state'} eq "Cancelled") {
1316 Log (undef, "Job cancelled at " . $Job->{'cancelled_at'} . " by user " . $Job->{'cancelled_by_user_uuid'});
1318 Log (undef, "Job state unexpectedly changed to " . $Job->{'state'});
1321 $main::please_freeze = 1;
1328 my $last_squeue_check = $squeue_checked;
1330 # Do not call `squeue` or check the kill list more than once every
1332 return if $last_squeue_check > time - 15;
1333 $squeue_checked = time;
1335 # Look for children from which we haven't received stderr data since
1336 # the last squeue check. If no such children exist, all procs are
1337 # alive and there's no need to even look at squeue.
1339 # As long as the crunchstat poll interval (10s) is shorter than the
1340 # squeue check interval (15s) this should make the squeue check an
1342 my $silent_procs = 0;
1343 for my $jobstep (values %proc)
1345 if ($jobstep->{stderr_at} < $last_squeue_check)
1350 return if $silent_procs == 0;
1352 # use killem() on procs whose killtime is reached
1353 while (my ($pid, $jobstep) = each %proc)
1355 if (exists $jobstep->{killtime}
1356 && $jobstep->{killtime} <= time
1357 && $jobstep->{stderr_at} < $last_squeue_check)
1360 if ($jobstep->{stderr_at}) {
1361 $sincewhen = " in last " . (time - $jobstep->{stderr_at}) . "s";
1363 Log($jobstep->{jobstep}, "killing orphaned srun process $pid (task not in slurm queue, no stderr received$sincewhen)");
1370 # here is an opportunity to check for mysterious problems with local procs
1374 # Get a list of steps still running. Note: squeue(1) says --steps
1375 # selects a format (which we override anyway) and allows us to
1376 # specify which steps we're interested in (which we don't).
1377 # Importantly, it also changes the meaning of %j from "job name" to
1378 # "step name" and (although this isn't mentioned explicitly in the
1379 # docs) switches from "one line per job" mode to "one line per step"
1380 # mode. Without it, we'd just get a list of one job, instead of a
1382 my @squeue = `squeue --jobs=\Q$ENV{SLURM_JOB_ID}\E --steps --format='%j' --noheader`;
1385 Log(undef, "warning: squeue exit status $? ($!)");
1390 # which of my jobsteps are running, according to squeue?
1392 for my $jobstepname (@squeue)
1394 $ok{$jobstepname} = 1;
1397 # Check for child procs >60s old and not mentioned by squeue.
1398 while (my ($pid, $jobstep) = each %proc)
1400 if ($jobstep->{time} < time - 60
1401 && $jobstep->{jobstepname}
1402 && !exists $ok{$jobstep->{jobstepname}}
1403 && !exists $jobstep->{killtime})
1405 # According to slurm, this task has ended (successfully or not)
1406 # -- but our srun child hasn't exited. First we must wait (30
1407 # seconds) in case this is just a race between communication
1408 # channels. Then, if our srun child process still hasn't
1409 # terminated, we'll conclude some slurm communication
1410 # error/delay has caused the task to die without notifying srun,
1411 # and we'll kill srun ourselves.
1412 $jobstep->{killtime} = time + 30;
1413 Log($jobstep->{jobstep}, "notice: task is not in slurm queue but srun process $pid has not exited");
1419 sub release_allocation
1423 Log (undef, "release job allocation");
1424 system "scancel $ENV{SLURM_JOB_ID}";
1432 foreach my $job (keys %reader)
1435 while (0 < sysread ($reader{$job}, $buf, 8192))
1437 print STDERR $buf if $ENV{CRUNCH_DEBUG};
1438 $jobstep[$job]->{stderr_at} = time;
1439 $jobstep[$job]->{stderr} .= $buf;
1440 preprocess_stderr ($job);
1441 if (length ($jobstep[$job]->{stderr}) > 16384)
1443 substr ($jobstep[$job]->{stderr}, 0, 8192) = "";
1452 sub preprocess_stderr
1456 while ($jobstep[$job]->{stderr} =~ /^(.*?)\n/) {
1458 substr $jobstep[$job]->{stderr}, 0, 1+length($line), "";
1459 Log ($job, "stderr $line");
1460 if ($line =~ /srun: error: (SLURM job $ENV{SLURM_JOB_ID} has expired|Unable to confirm allocation for job $ENV{SLURM_JOB_ID})/) {
1462 $main::please_freeze = 1;
1464 elsif ($line =~ /srun: error: (Node failure on|Aborting, io error)/) {
1465 my $job_slot_index = $jobstep[$job]->{slotindex};
1466 $slot[$job_slot_index]->{node}->{fail_count}++;
1467 $jobstep[$job]->{tempfail} = 1;
1468 ban_node_by_slot($job_slot_index);
1470 elsif ($line =~ /srun: error: (Unable to create job step|.*: Communication connection failure)/) {
1471 $jobstep[$job]->{tempfail} = 1;
1472 ban_node_by_slot($jobstep[$job]->{slotindex});
1474 elsif ($line =~ /arvados\.errors\.Keep/) {
1475 $jobstep[$job]->{tempfail} = 1;
1484 my $task_success = shift;
1485 preprocess_stderr ($job);
1488 Log ($job, "stderr $_");
1489 } split ("\n", $jobstep[$job]->{stderr});
1496 if (!open($keep, "-|", "arv-get", "--retries", retry_count(), $hash)) {
1497 Log(undef, "fetch_block run error from arv-get $hash: $!");
1500 my $output_block = "";
1503 my $bytes = sysread($keep, $buf, 1024 * 1024);
1504 if (!defined $bytes) {
1505 Log(undef, "fetch_block read error from arv-get: $!");
1506 $output_block = undef;
1508 } elsif ($bytes == 0) {
1509 # sysread returns 0 at the end of the pipe.
1512 # some bytes were read into buf.
1513 $output_block .= $buf;
1518 Log(undef, "fetch_block arv-get exited " . exit_status_s($?));
1519 $output_block = undef;
1521 return $output_block;
1524 # Create a collection by concatenating the output of all tasks (each
1525 # task's output is either a manifest fragment, a locator for a
1526 # manifest fragment stored in Keep, or nothing at all). Return the
1527 # portable_data_hash of the new collection.
1528 sub create_output_collection
1530 Log (undef, "collate");
1532 my ($child_out, $child_in);
1533 my $pid = open2($child_out, $child_in, 'python', '-c', q{
1536 print (arvados.api("v1").collections().
1537 create(body={"manifest_text": sys.stdin.read()}).
1538 execute(num_retries=int(sys.argv[1]))["portable_data_hash"])
1542 my $manifest_size = 0;
1546 my $output = $_->{'arvados_task'}->{output};
1547 next if (!defined($output));
1549 if ($output =~ /^[0-9a-f]{32}(\+\S+)*$/) {
1550 $next_write = fetch_block($output);
1552 $next_write = $output;
1554 if (defined($next_write)) {
1555 if (!defined(syswrite($child_in, $next_write))) {
1556 # There's been an error writing. Stop the loop.
1557 # We'll log details about the exit code later.
1560 $manifest_size += length($next_write);
1563 my $uuid = $_->{'arvados_task'}->{'uuid'};
1564 Log (undef, "Error retrieving '$output' output by task $task_idx ($uuid)");
1569 Log(undef, "collated output manifest text to send to API server is $manifest_size bytes with access tokens");
1572 my $s = IO::Select->new($child_out);
1573 if ($s->can_read(120)) {
1574 sysread($child_out, $joboutput, 1024 * 1024);
1577 Log(undef, "output collection creation exited " . exit_status_s($?));
1583 Log (undef, "timed out while creating output collection");
1584 foreach my $signal (2, 2, 2, 15, 15, 9) {
1585 kill($signal, $pid);
1586 last if waitpid($pid, WNOHANG) == -1;
1595 # Calls create_output_collection, logs the result, and returns it.
1596 # If that was successful, save that as the output in the job record.
1597 sub save_output_collection {
1598 my $collated_output = create_output_collection();
1600 if (!$collated_output) {
1601 Log(undef, "Failed to write output collection");
1604 Log(undef, "job output $collated_output");
1605 $Job->update_attributes('output' => $collated_output);
1607 return $collated_output;
1614 my $sig = 2; # SIGINT first
1615 if (exists $proc{$_}->{"sent_$sig"} &&
1616 time - $proc{$_}->{"sent_$sig"} > 4)
1618 $sig = 15; # SIGTERM if SIGINT doesn't work
1620 if (exists $proc{$_}->{"sent_$sig"} &&
1621 time - $proc{$_}->{"sent_$sig"} > 4)
1623 $sig = 9; # SIGKILL if SIGTERM doesn't work
1625 if (!exists $proc{$_}->{"sent_$sig"})
1627 Log ($proc{$_}->{jobstep}, "sending 2x signal $sig to pid $_");
1629 select (undef, undef, undef, 0.1);
1632 kill $sig, $_; # srun wants two SIGINT to really interrupt
1634 $proc{$_}->{"sent_$sig"} = time;
1635 $proc{$_}->{"killedafter"} = time - $proc{$_}->{"time"};
1645 vec($bits,fileno($_),1) = 1;
1651 # Send log output to Keep via arv-put.
1653 # $log_pipe_in and $log_pipe_out are the input and output filehandles to the arv-put pipe.
1654 # $log_pipe_out_buf is a string containing all output read from arv-put so far.
1655 # $log_pipe_out_select is an IO::Select object around $log_pipe_out.
1656 # $log_pipe_pid is the pid of the arv-put subprocess.
1658 # The only functions that should access these variables directly are:
1660 # log_writer_start($logfilename)
1661 # Starts an arv-put pipe, reading data on stdin and writing it to
1662 # a $logfilename file in an output collection.
1664 # log_writer_read_output([$timeout])
1665 # Read output from $log_pipe_out and append it to $log_pipe_out_buf.
1666 # Passes $timeout to the select() call, with a default of 0.01.
1667 # Returns the result of the last read() call on $log_pipe_out, or
1668 # -1 if read() wasn't called because select() timed out.
1669 # Only other log_writer_* functions should need to call this.
1671 # log_writer_send($txt)
1672 # Writes $txt to the output log collection.
1674 # log_writer_finish()
1675 # Closes the arv-put pipe and returns the output that it produces.
1677 # log_writer_is_active()
1678 # Returns a true value if there is currently a live arv-put
1679 # process, false otherwise.
1681 my ($log_pipe_in, $log_pipe_out, $log_pipe_out_buf, $log_pipe_out_select,
1684 sub log_writer_start($)
1686 my $logfilename = shift;
1687 $log_pipe_pid = open2($log_pipe_out, $log_pipe_in,
1691 '--filename', $logfilename,
1693 $log_pipe_out_buf = "";
1694 $log_pipe_out_select = IO::Select->new($log_pipe_out);
1697 sub log_writer_read_output {
1698 my $timeout = shift || 0.01;
1700 while ($read && $log_pipe_out_select->can_read($timeout)) {
1701 $read = read($log_pipe_out, $log_pipe_out_buf, 65536,
1702 length($log_pipe_out_buf));
1704 if (!defined($read)) {
1705 Log(undef, "error reading log manifest from arv-put: $!");
1710 sub log_writer_send($)
1713 print $log_pipe_in $txt;
1714 log_writer_read_output();
1717 sub log_writer_finish()
1719 return unless $log_pipe_pid;
1721 close($log_pipe_in);
1723 my $logger_failed = 0;
1724 my $read_result = log_writer_read_output(120);
1725 if ($read_result == -1) {
1726 $logger_failed = -1;
1727 Log (undef, "timed out reading from 'arv-put'");
1728 } elsif ($read_result != 0) {
1729 $logger_failed = -2;
1730 Log(undef, "failed to read arv-put log manifest to EOF");
1733 waitpid($log_pipe_pid, 0);
1735 $logger_failed ||= $?;
1736 Log(undef, "log_writer_finish: arv-put exited " . exit_status_s($?))
1739 close($log_pipe_out);
1740 my $arv_put_output = $logger_failed ? undef : $log_pipe_out_buf;
1741 $log_pipe_pid = $log_pipe_in = $log_pipe_out = $log_pipe_out_buf =
1742 $log_pipe_out_select = undef;
1744 return $arv_put_output;
1747 sub log_writer_is_active() {
1748 return $log_pipe_pid;
1751 sub Log # ($jobstep_id, $logmessage)
1753 if ($_[1] =~ /\n/) {
1754 for my $line (split (/\n/, $_[1])) {
1759 my $fh = select STDERR; $|=1; select $fh;
1760 my $message = sprintf ("%s %d %s %s", $job_id, $$, @_);
1761 $message =~ s{([^ -\176])}{"\\" . sprintf ("%03o", ord($1))}ge;
1764 if (log_writer_is_active() || -t STDERR) {
1765 my @gmtime = gmtime;
1766 $datetime = sprintf ("%04d-%02d-%02d_%02d:%02d:%02d",
1767 $gmtime[5]+1900, $gmtime[4]+1, @gmtime[3,2,1,0]);
1769 print STDERR ((-t STDERR) ? ($datetime." ".$message) : $message);
1771 if (log_writer_is_active()) {
1772 log_writer_send($datetime . " " . $message);
1779 my ($package, $file, $line) = caller;
1780 my $message = "@_ at $file line $line\n";
1781 Log (undef, $message);
1782 freeze() if @jobstep_todo;
1783 create_output_collection() if @jobstep_todo;
1793 if ($Job->{'state'} eq 'Cancelled') {
1794 $Job->update_attributes('finished_at' => scalar gmtime);
1796 $Job->update_attributes('state' => 'Failed');
1803 my $justcheckpoint = shift; # false if this will be the last meta saved
1804 return if $justcheckpoint; # checkpointing is not relevant post-Warehouse.pm
1805 return unless log_writer_is_active();
1806 my $log_manifest = log_writer_finish();
1807 return unless defined($log_manifest);
1810 my $prev_log_coll = api_call("collections/get", uuid => $Job->{log});
1811 $log_manifest = $prev_log_coll->{manifest_text} . $log_manifest;
1814 my $log_coll = api_call(
1815 "collections/create", ensure_unique_name => 1, collection => {
1816 manifest_text => $log_manifest,
1817 owner_uuid => $Job->{owner_uuid},
1818 name => sprintf("Log from %s job %s", $Job->{script}, $Job->{uuid}),
1820 Log(undef, "log collection is " . $log_coll->{portable_data_hash});
1821 $Job->update_attributes('log' => $log_coll->{portable_data_hash});
1825 sub freeze_if_want_freeze
1827 if ($main::please_freeze)
1829 release_allocation();
1832 # kill some srun procs before freeze+stop
1833 map { $proc{$_} = {} } @_;
1836 killem (keys %proc);
1837 select (undef, undef, undef, 0.1);
1839 while (($died = waitpid (-1, WNOHANG)) > 0)
1841 delete $proc{$died};
1846 create_output_collection();
1856 Log (undef, "Freeze not implemented");
1863 croak ("Thaw not implemented");
1879 $s =~ s{\\(.)}{$1 eq "n" ? "\n" : $1}ge;
1886 my $srunargs = shift;
1887 my $execargs = shift;
1888 my $opts = shift || {};
1890 my $args = $have_slurm ? [@$srunargs, @$execargs] : $execargs;
1892 $Data::Dumper::Terse = 1;
1893 $Data::Dumper::Indent = 0;
1894 my $show_cmd = Dumper($args);
1895 $show_cmd =~ s/(TOKEN\\*=)[^\s\']+/${1}[...]/g;
1896 $show_cmd =~ s/\n/ /g;
1897 if ($opts->{fork}) {
1898 Log(undef, "starting: $show_cmd");
1900 # This is a child process: parent is in charge of reading our
1901 # stderr and copying it to Log() if needed.
1902 warn "starting: $show_cmd\n";
1905 if (defined $stdin) {
1906 my $child = open STDIN, "-|";
1907 defined $child or die "no fork: $!";
1909 print $stdin or die $!;
1910 close STDOUT or die $!;
1915 return system (@$args) if $opts->{fork};
1918 warn "ENV size is ".length(join(" ",%ENV));
1919 die "exec failed: $!: @$args";
1923 sub ban_node_by_slot {
1924 # Don't start any new jobsteps on this node for 60 seconds
1926 $slot[$slotid]->{node}->{hold_until} = 60 + scalar time;
1927 $slot[$slotid]->{node}->{hold_count}++;
1928 Log (undef, "backing off node " . $slot[$slotid]->{node}->{name} . " for 60 seconds");
1933 my ($lockfile, $error_message) = @_;
1934 open L, ">", $lockfile or croak("$lockfile: $!");
1935 if (!flock L, LOCK_EX|LOCK_NB) {
1936 croak("Can't lock $lockfile: $error_message\n");
1940 sub find_docker_image {
1941 # Given a Keep locator, check to see if it contains a Docker image.
1942 # If so, return its stream name and Docker hash.
1943 # If not, return undef for both values.
1944 my $locator = shift;
1945 my ($streamname, $filename);
1946 my $image = api_call("collections/get", uuid => $locator);
1948 foreach my $line (split(/\n/, $image->{manifest_text})) {
1949 my @tokens = split(/\s+/, $line);
1951 $streamname = shift(@tokens);
1952 foreach my $filedata (grep(/^\d+:\d+:/, @tokens)) {
1953 if (defined($filename)) {
1954 return (undef, undef); # More than one file in the Collection.
1956 $filename = (split(/:/, $filedata, 3))[2];
1961 if (defined($filename) and ($filename =~ /^([0-9A-Fa-f]{64})\.tar$/)) {
1962 return ($streamname, $1);
1964 return (undef, undef);
1969 # Calculate the number of times an operation should be retried,
1970 # assuming exponential backoff, and that we're willing to retry as
1971 # long as tasks have been running. Enforce a minimum of 3 retries.
1972 my ($starttime, $endtime, $timediff, $retries);
1974 $starttime = $jobstep[0]->{starttime};
1975 $endtime = $jobstep[-1]->{finishtime};
1977 if (!defined($starttime)) {
1979 } elsif (!defined($endtime)) {
1980 $timediff = time - $starttime;
1982 $timediff = ($endtime - $starttime) - (time - $endtime);
1984 if ($timediff > 0) {
1985 $retries = int(log($timediff) / log(2));
1987 $retries = 1; # Use the minimum.
1989 return ($retries > 3) ? $retries : 3;
1993 # Pass in two function references.
1994 # This method will be called with the remaining arguments.
1995 # If it dies, retry it with exponential backoff until it succeeds,
1996 # or until the current retry_count is exhausted. After each failure
1997 # that can be retried, the second function will be called with
1998 # the current try count (0-based), next try time, and error message.
1999 my $operation = shift;
2000 my $retry_callback = shift;
2001 my $retries = retry_count();
2002 foreach my $try_count (0..$retries) {
2003 my $next_try = time + (2 ** $try_count);
2004 my $result = eval { $operation->(@_); };
2007 } elsif ($try_count < $retries) {
2008 $retry_callback->($try_count, $next_try, $@);
2009 my $sleep_time = $next_try - time;
2010 sleep($sleep_time) if ($sleep_time > 0);
2013 # Ensure the error message ends in a newline, so Perl doesn't add
2014 # retry_op's line number to it.
2020 # Pass in a /-separated API method name, and arguments for it.
2021 # This function will call that method, retrying as needed until
2022 # the current retry_count is exhausted, with a log on the first failure.
2023 my $method_name = shift;
2024 my $log_api_retry = sub {
2025 my ($try_count, $next_try_at, $errmsg) = @_;
2026 $errmsg =~ s/\s*\bat \Q$0\E line \d+\.?\s*//;
2027 $errmsg =~ s/\s/ /g;
2028 $errmsg =~ s/\s+$//;
2030 if ($next_try_at < time) {
2031 $retry_msg = "Retrying.";
2033 my $next_try_fmt = strftime "%Y-%m-%dT%H:%M:%SZ", gmtime($next_try_at);
2034 $retry_msg = "Retrying at $next_try_fmt.";
2036 Log(undef, "API method $method_name failed: $errmsg. $retry_msg");
2039 foreach my $key (split(/\//, $method_name)) {
2040 $method = $method->{$key};
2042 return retry_op(sub { $method->execute(@_); }, $log_api_retry, @_);
2046 # Given a $?, return a human-readable exit code string like "0" or
2047 # "1" or "0 with signal 1" or "1 with signal 11".
2048 my $exitcode = shift;
2049 my $s = $exitcode >> 8;
2050 if ($exitcode & 0x7f) {
2051 $s .= " with signal " . ($exitcode & 0x7f);
2053 if ($exitcode & 0x80) {
2054 $s .= " with core dump";
2059 sub handle_readall {
2060 # Pass in a glob reference to a file handle.
2061 # Read all its contents and return them as a string.
2062 my $fh_glob_ref = shift;
2064 return <$fh_glob_ref>;
2067 sub tar_filename_n {
2069 return sprintf("%s/git.%s.%d.tar", $ENV{CRUNCH_TMP}, $job_id, $n);
2072 sub add_git_archive {
2073 # Pass in a git archive command as a string or list, a la system().
2074 # This method will save its output to be included in the archive sent to the
2078 if (!open(GIT_ARCHIVE, ">", tar_filename_n($git_tar_count))) {
2079 croak("Failed to save git archive: $!");
2081 my $git_pid = open2(">&GIT_ARCHIVE", $git_input, @_);
2083 waitpid($git_pid, 0);
2086 croak("Failed to save git archive: git exited " . exit_status_s($?));
2090 sub combined_git_archive {
2091 # Combine all saved tar archives into a single archive, then return its
2092 # contents in a string. Return undef if no archives have been saved.
2093 if ($git_tar_count < 1) {
2096 my $base_tar_name = tar_filename_n(1);
2097 foreach my $tar_to_append (map { tar_filename_n($_); } (2..$git_tar_count)) {
2098 my $tar_exit = system("tar", "-Af", $base_tar_name, $tar_to_append);
2099 if ($tar_exit != 0) {
2100 croak("Error preparing build archive: tar -A exited " .
2101 exit_status_s($tar_exit));
2104 if (!open(GIT_TAR, "<", $base_tar_name)) {
2105 croak("Could not open build archive: $!");
2107 my $tar_contents = handle_readall(\*GIT_TAR);
2109 return $tar_contents;
2112 sub set_nonblocking {
2114 my $flags = fcntl ($fh, F_GETFL, 0) or croak ($!);
2115 fcntl ($fh, F_SETFL, $flags | O_NONBLOCK) or croak ($!);
2121 # This is crunch-job's internal dispatch script. crunch-job running on the API
2122 # server invokes this script on individual compute nodes, or localhost if we're
2123 # running a job locally. It gets called in two modes:
2125 # * No arguments: Installation mode. Read a tar archive from the DATA
2126 # file handle; it includes the Crunch script's source code, and
2127 # maybe SDKs as well. Those should be installed in the proper
2128 # locations. This runs outside of any Docker container, so don't try to
2129 # introspect Crunch's runtime environment.
2131 # * With arguments: Crunch script run mode. This script should set up the
2132 # environment, then run the command specified in the arguments. This runs
2133 # inside any Docker container.
2136 use File::Path qw( make_path remove_tree );
2137 use POSIX qw(getcwd);
2139 use constant TASK_TEMPFAIL => 111;
2141 # Map SDK subdirectories to the path environments they belong to.
2142 my %SDK_ENVVARS = ("perl/lib" => "PERLLIB", "ruby/lib" => "RUBYLIB");
2144 my $destdir = $ENV{"CRUNCH_SRC"};
2145 my $archive_hash = $ENV{"CRUNCH_GIT_ARCHIVE_HASH"};
2146 my $repo = $ENV{"CRUNCH_SRC_URL"};
2147 my $install_dir = $ENV{"CRUNCH_INSTALL"} || (getcwd() . "/opt");
2148 my $job_work = $ENV{"JOB_WORK"};
2149 my $task_work = $ENV{"TASK_WORK"};
2151 open(STDOUT_ORIG, ">&", STDOUT);
2152 open(STDERR_ORIG, ">&", STDERR);
2154 for my $dir ($destdir, $job_work, $task_work) {
2157 -e $dir or die "Failed to create temporary directory ($dir): $!";
2162 remove_tree($task_work, {keep_root => 1});
2165 ### Crunch script run mode
2167 # We want to do routine logging during task 0 only. This gives the user
2168 # the information they need, but avoids repeating the information for every
2171 if ($ENV{TASK_SEQUENCE} eq "0") {
2174 printf STDERR_ORIG "[Crunch] $msg\n", @_;
2180 my $python_src = "$install_dir/python";
2181 my $venv_dir = "$job_work/.arvados.venv";
2182 my $venv_built = -e "$venv_dir/bin/activate";
2183 if ((!$venv_built) and (-d $python_src) and can_run("virtualenv")) {
2184 shell_or_die(undef, "virtualenv", "--quiet", "--system-site-packages",
2185 "--python=python2.7", $venv_dir);
2186 shell_or_die(TASK_TEMPFAIL, "$venv_dir/bin/pip", "--quiet", "install", "-I", $python_src);
2188 $Log->("Built Python SDK virtualenv");
2191 my @pysdk_version_cmd = ("python", "-c",
2192 "from pkg_resources import get_distribution as get; print get('arvados-python-client').version");
2194 $Log->("Running in Python SDK virtualenv");
2195 @pysdk_version_cmd = ();
2196 my $orig_argv = join(" ", map { quotemeta($_); } @ARGV);
2197 @ARGV = ("/bin/sh", "-ec",
2198 ". \Q$venv_dir/bin/activate\E; exec $orig_argv");
2199 } elsif (-d $python_src) {
2200 $Log->("Warning: virtualenv not found inside Docker container default " .
2201 "\$PATH. Can't install Python SDK.");
2204 if (@pysdk_version_cmd) {
2205 open(my $pysdk_version_pipe, "-|", @pysdk_version_cmd);
2206 my $pysdk_version = <$pysdk_version_pipe>;
2207 close($pysdk_version_pipe);
2209 chomp($pysdk_version);
2210 $Log->("Using Arvados SDK version $pysdk_version");
2212 # A lot could've gone wrong here, but pretty much all of it means that
2213 # Python won't be able to load the Arvados SDK.
2214 $Log->("Warning: Arvados SDK not found");
2218 while (my ($sdk_dir, $sdk_envkey) = each(%SDK_ENVVARS)) {
2219 my $sdk_path = "$install_dir/$sdk_dir";
2221 if ($ENV{$sdk_envkey}) {
2222 $ENV{$sdk_envkey} = "$sdk_path:" . $ENV{$sdk_envkey};
2224 $ENV{$sdk_envkey} = $sdk_path;
2226 $Log->("Arvados SDK added to %s", $sdk_envkey);
2231 die "Cannot exec `@ARGV`: $!";
2234 ### Installation mode
2235 open L, ">", "$destdir.lock" or die "$destdir.lock: $!";
2237 if (readlink ("$destdir.archive_hash") eq $archive_hash && -d $destdir) {
2238 # This exact git archive (source + arvados sdk) is already installed
2239 # here, so there's no need to reinstall it.
2241 # We must consume our DATA section, though: otherwise the process
2242 # feeding it to us will get SIGPIPE.
2244 while (read(DATA, $buf, 65536)) { }
2249 unlink "$destdir.archive_hash";
2253 # Ignore SIGPIPE: we check retval of close() instead. See perlipc(1).
2254 local $SIG{PIPE} = "IGNORE";
2255 warn "Extracting archive: $archive_hash\n";
2256 # --ignore-zeros is necessary sometimes: depending on how much NUL
2257 # padding tar -A put on our combined archive (which in turn depends
2258 # on the length of the component archives) tar without
2259 # --ignore-zeros will exit before consuming stdin and cause close()
2260 # to fail on the resulting SIGPIPE.
2261 if (!open(TARX, "|-", "tar", "--ignore-zeros", "-xC", $destdir)) {
2262 die "Error launching 'tar -xC $destdir': $!";
2264 # If we send too much data to tar in one write (> 4-5 MiB), it stops, and we
2265 # get SIGPIPE. We must feed it data incrementally.
2267 while (read(DATA, $tar_input, 65536)) {
2268 print TARX $tar_input;
2271 die "'tar -xC $destdir' exited $?: $!";
2277 my $sdk_root = "$destdir/.arvados.sdk/sdk";
2279 foreach my $sdk_lang (("python",
2280 map { (split /\//, $_, 2)[0]; } keys(%SDK_ENVVARS))) {
2281 if (-d "$sdk_root/$sdk_lang") {
2282 if (!rename("$sdk_root/$sdk_lang", "$install_dir/$sdk_lang")) {
2283 die "Failed to install $sdk_lang SDK: $!";
2289 my $python_dir = "$install_dir/python";
2290 if ((-d $python_dir) and can_run("python2.7")) {
2291 open(my $egg_info_pipe, "-|",
2292 "python2.7 \Q$python_dir/setup.py\E egg_info 2>&1 >/dev/null");
2293 my @egg_info_errors = <$egg_info_pipe>;
2294 close($egg_info_pipe);
2297 if (@egg_info_errors and (($egg_info_errors[-1] =~ /\bgit\b/) or ($egg_info_errors[-1] =~ /\[Errno 2\]/))) {
2298 # egg_info apparently failed because it couldn't ask git for a build tag.
2299 # Specify no build tag.
2300 open(my $pysdk_cfg, ">>", "$python_dir/setup.cfg");
2301 print $pysdk_cfg "\n[egg_info]\ntag_build =\n";
2304 my $egg_info_exit = $? >> 8;
2305 foreach my $errline (@egg_info_errors) {
2308 warn "python setup.py egg_info failed: exit $egg_info_exit";
2309 exit ($egg_info_exit || 1);
2314 # Hide messages from the install script (unless it fails: shell_or_die
2315 # will show $destdir.log in that case).
2316 open(STDOUT, ">>", "$destdir.log");
2317 open(STDERR, ">&", STDOUT);
2319 if (-e "$destdir/crunch_scripts/install") {
2320 shell_or_die (undef, "$destdir/crunch_scripts/install", $install_dir);
2321 } elsif (!-e "./install.sh" && -e "./tests/autotests.sh") {
2323 shell_or_die (undef, "./tests/autotests.sh", $install_dir);
2324 } elsif (-e "./install.sh") {
2325 shell_or_die (undef, "./install.sh", $install_dir);
2328 if ($archive_hash) {
2329 unlink "$destdir.archive_hash.new";
2330 symlink ($archive_hash, "$destdir.archive_hash.new") or die "$destdir.archive_hash.new: $!";
2331 rename ("$destdir.archive_hash.new", "$destdir.archive_hash") or die "$destdir.archive_hash: $!";
2337 my $command_name = shift;
2338 open(my $which, "-|", "which", $command_name);
2339 while (<$which>) { }
2346 my $exitcode = shift;
2348 if ($ENV{"DEBUG"}) {
2349 print STDERR "@_\n";
2351 if (system (@_) != 0) {
2354 my $exitstatus = sprintf("exit %d signal %d", $code >> 8, $code & 0x7f);
2355 open STDERR, ">&STDERR_ORIG";
2356 system ("cat $destdir.log >&2");
2357 warn "@_ failed ($err): $exitstatus";
2358 if (defined($exitcode)) {
2362 exit (($code >> 8) || 1);