#!/usr/bin/env expect
############################################################################
# Purpose: Test of Slurm functionality
#          Test that job's node estimation is based off the biggest node
#          in the partition
############################################################################
# Copyright (C) 2015 SchedMD LLC
# Written by Nathan Yee <nyee32@schedmd.com>
#
# 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 node_list        ""
set highest_cpu_cnt  0
set default_part     [default_partition]
set script           "$test_dir/job_script"
set job_id           0
set node_l           ""
set cpu_l_cnt        0
set node_m           ""
set cpu_m_cnt        0
set node_h           ""
set cpu_h_cnt        0
set test_part        "${test_name}_part"

if {![is_super_user]} {
    skip "This test can't be run without being a super user of the cluster"
} elseif {[param_contains [get_config_param "SelectTypeParameters"] "CR_ONE_TASK_PER_CORE"]} {
	skip "This test is incompatible SelectTypeParameters=CR_ONE_TASK_PER_CORE"
}

proc cleanup {} {
	global job_id scontrol test_part

	cancel_job $job_id
	run_command "$scontrol delete partition=$test_part"
}

proc sub_job {cpu_cnt part} {
	global script job_id

	set job_id [submit_job -fail "-t1 -H -p$part -n$cpu_cnt -o/dev/null $script"]
}

proc check_node_cnt { exp_nodes } {
	global squeue

	set match 0
	spawn $squeue -h -o%D
	expect {
		-re "$exp_nodes" {
			set match 1
			exp_continue
		}
		timeout {
			fail "squeue is not responding"
		}
		eof {
			wait
		}
	}

	subtest {$match} "Job should make proper node estimation" "expected = $exp_nodes"
}

proc update_job { job_id part } {
	global scontrol

	spawn $scontrol update jobid=$job_id partition=$part
	expect {
		timeout {
			fail "scontrol is not responding"
		}
		eof {
			wait
		}
	}
}

proc test_and_check { job_id part exp_nodes } {
	update_job $job_id $part
	check_node_cnt $exp_nodes
}

make_bash_script $script "sleep 10"

# only works when job is in hold. should also work when jobs are waiting

################## Test with existing default partition ##################
spawn $bin_bash -c "$scontrol show partition $default_part | $bin_grep -w Nodes"
expect {
	-re "Nodes=($re_word_str)" {
		set node_list $expect_out(1,string)
		exp_continue
	}
	timeout {
		fail "scontrol is not responding"
	}
	eof {
		wait
	}
}

set node_cnt 0
spawn $bin_bash -c "$scontrol show nodes $node_list | grep CPUTot"
expect {
	-re "CPUTot=($number)" {
		incr node_cnt
		if { $expect_out(1,string) > $highest_cpu_cnt } {
			set highest_cpu_cnt $expect_out(1,string)
		}
		exp_continue
	}
	timeout {
		fail "scontrol is not responding"
	}
	eof {
		wait
	}
}

# Choose a CPU count equal to the highest CPU node (job should use 1 node)
set cpu_test_cnt $highest_cpu_cnt
sub_job $cpu_test_cnt $default_part
set exp_nodes [expr ($cpu_test_cnt - 1)/$highest_cpu_cnt + 1]
check_node_cnt $exp_nodes
cancel_job $job_id

# Choose a CPU count greater then highest CPU node (job should use 2 nodes)
if {$node_cnt >= 2} {
	set cpu_test_cnt [expr $highest_cpu_cnt + 1]
	sub_job $cpu_test_cnt $default_part
	set exp_nodes [expr ($cpu_test_cnt - 1)/$highest_cpu_cnt + 1]
	check_node_cnt $exp_nodes
	cancel_job $job_id
}

# Choose a CPU count greater then highest CPU node (job should use 5 nodes)
if {$node_cnt >= 5} {
	set cpu_test_cnt [expr $highest_cpu_cnt * 4 + 1]
	sub_job $cpu_test_cnt $default_part
	set exp_nodes [expr ($cpu_test_cnt - 1)/$highest_cpu_cnt + 1]
	check_node_cnt $exp_nodes
	cancel_job $job_id
}


################## Test with new partition ##################

# Get the smallest node in the system
log_user 0
set tmp_cnt 99999
set tmp_name ""
set node_name ""
spawn $sinfo -h -o%N=%c -N
expect {
	-re "($re_word_str)=($number)" {
		if { $expect_out(2,string) < $tmp_cnt } {
			set tmp_cnt $expect_out(2,string)
			set node_name $expect_out(1,string)
		}
		exp_continue

	}
	timeout {
		fail "scontrol is not responding"
	}
	eof {
		wait
	}
}

# Set node to node with least CPUs
set node_l $node_name
set cpu_l_cnt $tmp_cnt
set node_m $node_name
set cpu_m_cnt $tmp_cnt
set node_h $node_name
set cpu_h_cnt $tmp_cnt

# Get highest node of different
spawn $sinfo -h -o%N=%c -N
expect {
	-re "($re_word_str)=($number)" {
		if { $expect_out(2,string) > $cpu_l_cnt &&
		     $expect_out(2,string) > $cpu_h_cnt } {
			set cpu_h_cnt $expect_out(2,string)
			set node_h $expect_out(1,string)
		}
		exp_continue
	}
	timeout {
		fail "sinfo is not responding"
	}
	eof {
		wait
	}
}

# Get node between highest and lowest
spawn $sinfo -h -o%N=%c -N
expect {
	-re "($re_word_str)=($number)" {
		if { $expect_out(2,string) > $cpu_l_cnt &&
		     $expect_out(2,string) < $cpu_h_cnt } {
			set cpu_m_cnt $expect_out(2,string)
			set node_m $expect_out(1,string)
		}
		exp_continue
	}
	timeout {
		fail "sinfo is not responding"
	}
	eof {
		wait
	}
}
log_user 1

log_debug "L:$node_l:$cpu_l_cnt M:$node_m:$cpu_m_cnt H:$node_h:$cpu_h_cnt"
if {$cpu_l_cnt == $cpu_h_cnt} {
	log_debug "The rest of this test expects to have three different nodes \
		each with different cpu counts -- finishing test now."
	pass
}

# Create partition with the smallest node in the system
spawn $scontrol create partitionname=$test_part nodes=$node_l
expect {
	timeout {
		fail "scontrol is not responding"
	}
	eof {
		wait
	}
}

set match 0
spawn $sinfo -h -o%P
expect {
	-re "$test_part" {
		set match 1
		exp_continue
	}
	timeout {
		fail "scontrol is not responding"
	}
	eof {
		wait
	}
}

if {!$match} {
	fail "Partition ($test_part) was not created"
}

# Choose a cpu count greater then highest cpu node
set cpu_test_cnt [expr $cpu_l_cnt + $cpu_m_cnt + $cpu_h_cnt]
set exp_nodes [expr ($cpu_test_cnt - 1)/$cpu_l_cnt + 1]

sub_job $cpu_test_cnt $default_part
sleep 4
test_and_check $job_id $test_part $exp_nodes

sleep 4
cancel_job $job_id

# Update the partition with a new node
spawn $scontrol update partitionname=$test_part nodes=$node_l,$node_m
expect {
	timeout {
		fail "scontrol is not responding"
	}
	eof {
		wait
	}
}

set exp_nodes [expr ($cpu_test_cnt - 1)/$cpu_m_cnt + 1]

sub_job $cpu_test_cnt $default_part
sleep 4
test_and_check $job_id $test_part $exp_nodes

sleep 4
cancel_job $job_id

# Update the partition with a new node
spawn $scontrol update partitionname=$test_part nodes=$node_l,$node_m,$node_h
expect {
	timeout {
		fail "scontrol is not responding"
	}
	eof {
		wait
	}
}

set exp_nodes [expr ($cpu_test_cnt - 1)/$cpu_h_cnt + 1]

sub_job $cpu_test_cnt $default_part
sleep 4
test_and_check $job_id $test_part $exp_nodes