| #!/usr/bin/env expect |
| ############################################################################ |
| # Purpose: Test of Slurm functionality |
| # Validate that sacct -D shows correct job steps and states |
| # when a job is requeued |
| ############################################################################ |
| # Copyright (C) SchedMD LLC. |
| # |
| # This file is part of Slurm, a resource management program. |
| # For details, see <https://slurm.schedmd.com/>. |
| # Please also read the included file: DISCLAIMER. |
| # |
| # Slurm is free software; you can redistribute it and/or modify it under |
| # the terms of the GNU General Public License as published by the Free |
| # Software Foundation; either version 2 of the License, or (at your option) |
| # any later version. |
| # |
| # Slurm is distributed in the hope that it will be useful, but WITHOUT ANY |
| # WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS |
| # FOR A PARTICULAR PURPOSE. See the GNU General Public License for more |
| # details. |
| # |
| # You should have received a copy of the GNU General Public License along |
| # with Slurm; if not, write to the Free Software Foundation, Inc. |
| # 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. |
| ############################################################################ |
| source ./globals |
| |
| set job_id 0 |
| set node [get_nodes_by_request -fail "-t1 --exclusive"] |
| |
| set accounting_storage_enforce [get_config_param "AccountingStorageEnforce"] |
| if {[param_contains $accounting_storage_enforce "nosteps"] || [param_contains $accounting_storage_enforce "nojobs"]} { |
| skip "This test can not be run with nosteps or nojobs (AccountingStorageEnforce)" |
| } |
| if {![is_super_user]} { |
| skip "Test can only be run as SlurmUser" |
| } |
| |
| proc cleanup {} { |
| global job_id |
| |
| cancel_job $job_id |
| } |
| |
| proc mod_state { state reason } { |
| global scontrol node |
| |
| set output [run_command_output "$scontrol update nodename=$node state=$state reason=$reason"] |
| set bad_state [regexp "Invalid node state" $output] |
| |
| if {$bad_state == 1 && $state eq "resume" && [get_config_param "ReturnToService"] == 2} { |
| log_warn "This error is expected, no worries" |
| set bad_state 0 |
| } |
| if {$bad_state == 1} { |
| fail "Problem changing node state" |
| } |
| } |
| |
| proc check_step { num } { |
| global sacct job_id |
| |
| set steps -1 |
| wait_for {$num == $steps} { |
| set output [run_command_output -fail "$sacct --job=$job_id\.batch -D --start=now-15minutes --noheader --format=jobid -P"] |
| set steps [regexp -all "batch" $output] |
| } |
| subtest {$num == $steps} "Check number of steps" "$steps != $num" |
| } |
| |
| |
| if {[get_config_param "AccountingStorageType"] ne "accounting_storage/slurmdbd"} { |
| skip "Not using accounting_storage/slurmdbd" |
| } |
| |
| # Submit job to be requeued |
| log_info "Test 1" |
| set job_id [submit_job -fail "-N1 -w$node --exclusive -o/dev/null --requeue --wrap='$bin_sleep 20'"] |
| |
| wait_for_job -fail $job_id "RUNNING" |
| |
| # Wait for batch script to start (after message delays, prologs, etc.) |
| wait_for_step "${job_id}.batch" |
| |
| # Set the node that the job is running on to down |
| mod_state "down" "$test_name" |
| |
| # Wait a little bit for node state to change |
| set node_state_found 0 |
| wait_for -fail {$node_state_found} { |
| set nodes [get_nodes] |
| if {[dict get [dict get $nodes $node] "State"] == "DOWN"} { |
| set node_state_found 1 |
| } |
| } |
| |
| # Set the node back to resume |
| mod_state "resume" "$test_name" |
| |
| # Check the job state |
| log_info "Test 2" |
| wait_for_job -fail $job_id "PENDING" |
| |
| # Wait for the state changes to propagate to the database for sacct |
| set fail_count 0 |
| set canc_count 0 |
| set pend_count 0 |
| wait_for {$fail_count == 1 && $canc_count == 1 && $pend_count == 1} { |
| set output [run_command_output -fail "$sacct --job=$job_id --duplicates --parsable2 --start=now-15minutes --end=tomorrow --noheader -o JobID,State"] |
| |
| # Skip "extern" job container optionally spawned by "PrologFlags=contain" |
| set fail_count [regexp -all "\[0-9_\]+(\.(?!extern)\[a-z\]+)*\\|NODE_FAIL" $output] |
| set canc_count [regexp -all "\[0-9_\]+(\.(?!extern)\[a-z\]+)*\\|CANCELLED" $output] |
| set pend_count [regexp -all "\[0-9_\]+(\.(?!extern)\[a-z\]+)*\\|PENDING" $output] |
| } |
| subtest {$fail_count == 1} "Test NODE_FAIL count" "$fail_count != 1" |
| subtest {$canc_count == 1} "Test CANCELLED count" "$canc_count != 1" |
| subtest {$pend_count == 1} "Test PENDING count" "$pend_count != 1" |
| |
| wait_for_job -fail $job_id "RUNNING" |
| |
| # Wait for batch script to start (after message delays, prologs, etc.) |
| |
| log_info "Test 3" |
| set fail_count 0 |
| set canc_count 0 |
| set run_count 0 |
| wait_for {$fail_count == 1 && $canc_count == 1 && $run_count == 2} { |
| set output [run_command_output -fail "$sacct --job=$job_id --duplicates --parsable2 --start=now-15minutes --end=tomorrow --noheader -o JobID,State"] |
| |
| # Skip "extern" job container optionally spawned by "PrologFlags=contain" |
| set fail_count [regexp -all "\[0-9_\]+(\.(?!extern)\[a-z\]+)*\\|NODE_FAIL" $output] |
| set canc_count [regexp -all "\[0-9_\]+(\.(?!extern)\[a-z\]+)*\\|CANCELLED" $output] |
| set run_count [regexp -all "\[0-9_\]+(\.(?!extern)\[a-z\]+)*\\|RUNNING" $output] |
| } |
| subtest {$fail_count == 1} "Test NODE_FAIL count" "$fail_count != 1" |
| subtest {$canc_count == 1} "Test CANCELLED count" "$canc_count != 1" |
| subtest {$run_count == 2} "Test RUNNING count" "$run_count != 2" |
| |
| # Requeue the job |
| run_command -fail "$scontrol requeue $job_id" |
| |
| # Wait a bit for the job to be requeued then check its state |
| wait_for_job -fail $job_id "PENDING" |
| |
| # Wait for the state changes to propagate to the database for sacct |
| log_info "Test 4" |
| set fail_count 0 |
| set canc_count 0 |
| set pend_count 0 |
| set req_count 0 |
| wait_for {$fail_count == 1 && $canc_count == 2 && $pend_count == 1 && $req_count == 1} { |
| set output [run_command_output -fail "$sacct --job=$job_id --duplicates --parsable2 --start=now-15minutes --end=tomorrow --noheader -o JobID,State"] |
| |
| # Skip "extern" job container optionally spawned by "PrologFlags=contain" |
| set fail_count [regexp -all "\[0-9_\]+(\.(?!extern)\[a-z\]+)*\\|NODE_FAIL" $output] |
| set canc_count [regexp -all "\[0-9_\]+(\.(?!extern)\[a-z\]+)*\\|CANCELLED" $output] |
| set pend_count [regexp -all "\[0-9_\]+(\.(?!extern)\[a-z\]+)*\\|PENDING" $output] |
| set req_count [regexp -all "\[0-9_\]+(\.(?!extern)\[a-z\]+)*\\|REQUEUE" $output] |
| } |
| subtest {$fail_count == 1} "Test NODE_FAIL count" "$fail_count != 1" |
| subtest {$canc_count == 2} "The first and second batch steps should both show CANCELLED" "$canc_count != 2" |
| subtest {$pend_count == 1} "Test PENDING count" "$pend_count != 1" |
| subtest {$req_count == 1} "Test REQUEUE count" "$req_count != 1" |
| |
| wait_for_job -fail $job_id "RUNNING" |
| |
| # Check for steps after requeue. There should be 3 batch steps - the first 2 |
| # that are CANCELLED, and now the last one that is running. |
| check_step 3 |
| |
| |
| log_info "Test 5" |
| set fail_count 0 |
| set canc_count 0 |
| set req_count 0 |
| set run_count 0 |
| wait_for {$fail_count == 1 && $canc_count == 2 && $req_count == 1 && $run_count == 2} { |
| set output [run_command_output -fail "$sacct --job=$job_id --duplicates --parsable2 --start=now-15minutes --end=tomorrow --noheader -o JobID,State"] |
| |
| # Skip "extern" job container optionally spawned by "PrologFlags=contain" |
| set fail_count [regexp -all "\[0-9_\]+(\.(?!extern)\[a-z\]+)*\\|NODE_FAIL" $output] |
| set canc_count [regexp -all "\[0-9_\]+(\.(?!extern)\[a-z\]+)*\\|CANCELLED" $output] |
| set req_count [regexp -all "\[0-9_\]+(\.(?!extern)\[a-z\]+)*\\|REQUEUE" $output] |
| set run_count [regexp -all "\[0-9_\]+(\.(?!extern)\[a-z\]+)*\\|RUNNING" $output] |
| } |
| subtest {$fail_count == 1} "Test NODE_FAIL count" "$fail_count != 1" |
| subtest {$canc_count == 2} "Test CANCELLED count" "$canc_count != 2" |
| subtest {$req_count == 1} "Test REQUEUE count" "$req_count != 1" |
| subtest {$run_count == 2} "The job and its batch step should be RUNNING" "$run_count != 2" |
| |
| wait_for_job -fail $job_id "DONE" |
| |
| # Check steps after job has completed |
| check_step 3 |
| log_info "Test 6" |
| set fail_count 0 |
| set canc_count 0 |
| set comp_count 0 |
| set req_count 0 |
| wait_for {$fail_count == 1 && $canc_count == 2 && $comp_count == 2 && $req_count == 1} { |
| set output [run_command_output -fail "$sacct --job=$job_id --duplicates --parsable2 --start=now-15minutes --end=tomorrow --noheader -o JobID,State"] |
| |
| # Skip "extern" job container optionally spawned by "PrologFlags=contain" |
| set fail_count [regexp -all "\[0-9_\]+(\.(?!extern)\[a-z\]+)*\\|NODE_FAIL" $output] |
| set canc_count [regexp -all "\[0-9_\]+(\.(?!extern)\[a-z\]+)*\\|CANCELLED" $output] |
| set comp_count [regexp -all "\[0-9_\]+(\.(?!extern)\[a-z\]+)*\\|COMPLETED" $output] |
| set req_count [regexp -all "\[0-9_\]+(\.(?!extern)\[a-z\]+)*\\|REQUEUE" $output] |
| } |
| subtest {$fail_count == 1} "Test NODE_FAIL count" "$fail_count != 1" |
| subtest {$canc_count == 2} "Test CANCELLED count" "$canc_count != 2" |
| subtest {$comp_count == 2} "Test COMPLETED count" "$comp_count != 2" |
| subtest {$req_count == 1} "Test REQUEUE count" "$req_count != 1" |