#!/usr/bin/env expect
############################################################################
# Purpose: Test of Slurm functionality
#          Test that partition and job qos limits are enforced when using
#          the OverPartQos flag for the job's qos
#
# Output:  "TEST: #.#" followed by "SUCCESS" if test was successful, OR
#          "FAILURE: ..." otherwise with an explanation of the failure, OR
#          anything else indicates a failure mode that must be investigated.
############################################################################
# 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
source ./globals_accounting
source  ./inc21.34_tests

set test_id          21.34
set exit_code        0
set test_node        ""
# Total cpus in test node
set totcpus          0
set nthreads         0
set acct             test_acct
set user_name        ""
set test_part        "test$test_id\_part"
set part_qos         "test$test_id\_part_qos"
set job_qos          "test$test_id\_job_qos"
set qostest          ""
set grn              GrpNodes
set grn_num          0
set grcpu            GrpCpus
set grcpu_num        0
set grpcpumin        GrpCPUMins
set grpcpumin_num    0
# Set grpcpurunmin_num to multiple of CPUs per core to work with most configurations
# Also make sure that it is at least 4 so we can add and subtract from it
set grpcpurunmin     GrpCPURunMins
set grpcpurunmin_num 40
set grjobs           GrpJobs
set grjobs_num       2
set grpmem           GrpMem
set grpmem_num       100
set grsub            GrpSubmit
set grsub_num        2
set grpwall          GrpWall
set grpwall_num      1
set maxcpu           MaxCpus
set maxcpu_num       0
# Set maxcpumin_num to multiple of CPUs per core to work with most configurations
set maxcpumin        MaxCPUMins
set maxcpumin_num    2
set maxwall          MaxWall
set maxwall_num      2
set maxcpuspu        MaxCPUSPerUser
set maxcpuspu_num    2
set maxnodes         MaxNodes
set maxnode_num      0
set maxnodespu       MaxNodesPerUser
set maxnodespu_num   0
set maxjobs          MaxJobs
set maxjobs_num      2
set maxjobsub        MaxSubmitJobs
set maxjobsub_num    2
set time_spacing     1
set tres_cpu_mult    2

# cr_core = 1 / cr_cpu = 0
set selectparam      0
set def_part         [default_partition]

# mod qos
array set mod_job_qos {
	GrpNodes        -1
	GrpCpus         -1
	GrpJob          -1
	GrpSubmit       -1
	GrpCpuMin       -1
	GrpCpuRunMin    -1
	GrpMem          -1
	GrpWall         -1
	MaxCpus         -1
	MaxNode         -1
	MaxJobs         -1
	MaxSubmitJobs   -1
	MaxCpuMin       -1
	MaxWall         -1
	MaxCpusPerUser  -1
	MaxNode         -1
	MaxNodesPerUser -1

	GrpTRES=billing             -1
	GrpTRESMins=billing         -1
	GrpTRESRunMins=billing      -1
	MaxTRESPerJob=billing       -1
	MaxTRESMinsPerJob=billing   -1
	MaxTRESPerUser=billing      -1
}

array set mod_part_qos {
	GrpNodes        -1
	GrpCpus         -1
	GrpJob          -1
	GrpSubmit       -1
	GrpCpuMin       -1
	GrpCpuRunMin    -1
	GrpMem          -1
	GrpWall         -1
	MaxCpus         -1
	MaxNode         -1
	MaxJobs         -1
	MaxSubmitJobs   -1
	MaxCpuMin       -1
	MaxWall         -1
	MaxCpusPerUser  -1
	MaxNode         -1
	MaxNodesPerUser -1

	GrpTRES=billing             -1
	GrpTRESMins=billing         -1
	GrpTRESRunMins=billing      -1
	MaxTRESPerJob=billing       -1
	MaxTRESMinsPerJob=billing   -1
	MaxTRESPerUser=billing      -1
}

print_header $test_id

#
# Cannot run the test if OverTimeLimit is set, since we test time limits.
#
set overtimelim [get_over_time_limit]
if {$overtimelim != 0} {
	log_warn "Cannot run this test when OverTimeLimit is set. Exiting now."
	exit 0
}

proc cleanup { } {

	global acct job_qos part_qos scontrol sacctmgr test_part def_part
	global exit_code

	# Delete the test qos
	set match 0
	spawn $sacctmgr -i delete qos $job_qos,$part_qos
	expect {
		-re "Deleting" {
			set match 1
			exp_continue
		}
		timeout {
			send_user "\nFAILURE: sacctmgr is not responding\n"
			set exit_code 1
		}
		eof {
			wait
		}
	}

	#delete account
	spawn $sacctmgr -i delete account $acct
	expect {
		-re "Deleting accounts" {
			exp_continue
		}
		-re "Error" {
			send_user "\nFAILURE: account was not deleted\n"
			set exit_code 1
		}
		timeout {
			send_user "\nFAILURE: sacctmgr is not responding\n"
			set exit_code 1
		}
		eof {
			wait
		}
	}

	spawn $scontrol delete partitionname=$test_part
	expect {
		timeout {
			send_user "\nFAILURE: scontrol is not responding\n"
			set exit_code 1
		}
		eof {
			wait
		}
	}

	if {[string length $def_part]} {
		spawn $scontrol update partitionname=$def_part default=yes
		expect {
			timeout {
				send_user "\nFAILURE: scontrol is not responding\n"
				set exit_code 1
			}
			eof {
				wait
			}
		}
	}
}

if { [test_account_storage] == 0 } {
	send_user "\nWARNING: This test can't be run without a usable AccountStorageType\n"
	exit 0
} elseif { [test_enforce_limits] == 0 } {
	send_user "\nWARNING: This test can't be run without a usable AccountingStorageEnforce\n"
	exit 0
}
if { [test_limits_enforced] == 0 } {
	send_user "\nWARNING: This test can't be run without enforcing limits\n"
	exit 0
}
if {[test_super_user] == 0} {
	send_user "\nWARNING Test can only be ran as SlurmUser\n"
	exit 0
}
if {[test_select_type_params "CR_SOCKET"]} {
	send_user "\nWARNING: This test is incompatible with CR_SOCKET allocations\n"
	exit 0
}
if {[test_select_type_params "CR_ONE_TASK_PER_CORE"]} {
        send_user "\nWARNING: This test is incompatible with CR_ONE_TASK_PER_CORE allocations\n"
        exit 0
}

if { ![string compare [priority_type] "multifactor"] } {
	set prio_multifactor 1
} else {
	set prio_multifactor 0
}

# Remove any vestigial data
cleanup

# Check to see that there are enough resources in the default partition
set tmpc 0
set tmpn 0
spawn $scontrol show part [default_partition]
expect {
	-re "TotalCPUs=($number)" {
		set tmpc [expr $expect_out(1,string) - 1]
		exp_continue
	}
	-re "TotalNodes=($number)" {
		set tmpn [expr $expect_out(1,string) - 1]
		exp_continue
	}
	timeout {
		send_user "\nFAILURE: scontrol is not responding\n"
		set exit_code 1
	}
	eof {
		wait
	}
}

if {$tmpc == 0 || $tmpn == 0} {
	send_user "\nWARNING: not enough Nodes and/or CPUs\n"
	exit 0
}

# Determine what the selecttype param is
if {[test_select_type_params "CR_CORE"]} {
	set selectparam 1
}

# Get the number of nodes in the default partition
set num_nodes [available_nodes [default_partition] "idle"]

if {$num_nodes == 0} {
	send_user "\nFAILURE: no cpus where found\n"
	exit 1
} else {
	# Set QoS node values
	set grn_num     $num_nodes
	set maxnode_num $num_nodes
	set maxnodespu_num $num_nodes
}

# Create 2 test qos
add_qos $part_qos ""
add_qos $job_qos ""

# create a tmp partition to use for testing
spawn $scontrol create partitionname=$test_part qos=$part_qos tresbillingweights=cpu=$tres_cpu_mult default=yes \
    nodes=ALL
expect {
	timeout {
		send_user "\nFAILURE: scontrol is not responding\n"
		set exit_code
	}
	eof {
		wait
	}
}

set got_node 0
spawn $srun -N1 printenv SLURM_NODELIST
expect {
	-re "($alpha_numeric_under)" {
		set test_node $expect_out(1,string)
		set got_node 1
		exp_continue
	}
	timeout {
		send_user "\nFAILURE: srun is not responding\n"
		set exit_code 1
	}
	eof {
		wait
	}
}

if {$got_node != 1} {
	send_user "\nFAILURE: did not get node for testing\n"
	exit 0
}

# Get the number of cpus on a node
lassign [get_node_cpus $test_node] totcpus nthreads

if {$totcpus == 0} {
	send_user "\nFAILURE: no cpus where found\n"
	exit 1
} else {
	# Set QoS CPU values
	set grcpu_num [expr $totcpus - $nthreads]
	set grpcpumin_num $totcpus
	set maxcpu_num [expr $totcpus - $nthreads]
	set maxcpumin_num $totcpus
}

# Gets user
set user_name [get_my_user_name]

# Add account with qos
set acctmatch 0
spawn $sacctmgr -i add account $acct qos=$job_qos
expect {
	-re "Adding Account" {
		incr acctmatch
		exp_continue
	}
	timeout {
		send_user "\nFAILURE: sacctmgr is not responding\n"
		set exit_code 1
	}
	eof {
		wait

	}
}
if {$acctmatch != 1} {
	send_user "\nFAILURE: sacctmgr had a problem adding the account\n"
	cleanup
	exit 1
}

# Add user to account
spawn $sacctmgr -i create user name=$user_name account=$acct
expect {
	timeout {
		send_user "\nFAILURE: sacctmgr not responding\n"
	}
	eof {
		wait
	}
}


send_user "\n========== Run limit test on partition's qos limits ==========\n\n"
part_test

#
# Set overpartqos flag on job's qos
#
set changed 0
spawn $sacctmgr -i mod qos $job_qos set flag=overpartqos
expect {
	-re "Modified qos" {
		set changed 1
		exp_continue
	}
	timeout {
		send_user "\nFAILURE: sacctmgr is not resonding\n"
		set exit_code 1
	}
	eof {
		wait
	}
}


send_user "\n========== Run limit test on job's qos limits ==========\n\n"
qos_test

cleanup

if {$exit_code == 0} {
	print_success $test_id
} else {
	send_user "\nFAILURE: test $test_id\n"
}

exit $exit_code