Problem: Inspired by this thread, I'm trying to write a wrapper script that submits SLURM array jobs with bash variables. However, I'm running into issues with SLURM environment variables like $SLURM_ARRAY_TASK_ID as it acts as an empty variable.
I suspect it has something to do with how the test_wrapper.sh is parsing the yet undefined SLURM variable, but I can't seem to find a solution.
Below I provide a working example with a simple python script that should take an array ID as an input variable, but when it is called by the bash wrapper script, the python script crashes as it receives an empty variable.
test_wrapper.sh :
#!/bin/bash
for argument in "$#"
do
key=$(echo $argument | cut -f 1 -d'=')
value=$(echo $argument | cut -f 2 -d'=')
case "$key" in
"job_name") job_name="$value" ;;
"cpus") cpus="$value" ;;
"memory") memory="$value" ;;
"time") time="$value" ;;
"array") array="$value" ;;
*)
esac
done
sbatch <<EOT
#!/bin/bash
#SBATCH --account=foobar
#SBATCH --cpus-per-task=${cpus:-1}
#SBATCH --mem-per-cpu=${memory:-1}GB
#SBATCH --time=${time:-00:01:00}
#SBATCH --array=${array:-1-2}
#SBATCH --job-name=${job_name:-Default_Job_Name}
if [ -z "$SLURM_ARRAY_TASK_ID" ]
then
echo "The array ID \$SLURM_ARRAY_TASK_ID is empty"
else
echo "The array ID \$SLURM_ARRAY_TASK_ID is NOT empty"
fi
srun python foo.py -a $SLURM_ARRAY_TASK_ID
echo "Job finished with exit code $?"
EOT
where foo.py is:
import argparse
def main(args):
print('array number is : {}'.format(args.array_number))
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("-a", "--array_number",
help="the value passed from SLURM_ARRAY_TASK_ID"
)
args = parser.parse_args()
main(args)
$cat slurm-123456789_1.out yields :
The array ID 1 is empty
usage: foo.py [-h] [-a ARRAY_NUMBER]
foo.py: error: argument -a/--array_number: expected one argument
srun: error: nc10931: task 0: Exited with exit code 2
Job finished with exit code 0
I find it strange, that "The array ID 1 is empty" is correctly printing the $SLURM_ARRAY_TASK_ID (??)
So according to this page:
Job arrays will have two additional environment variable set. SLURM_ARRAY_JOB_ID will be set to the first job ID of the array. SLURM_ARRAY_TASK_ID will be set to the job array index value.
That suggests to me that sbatch is supposed to set these for you. In that case, you need to escape all instances of $SLURM_ARRAY_TASK_ID in the script you pass via the heredoc so that they don't get prematurely substituted before sbatch can set the relevant environment variable.
The two options for this are:
If you don't want any expansions to occur at all, quote the heredoc delimiter.
sbatch <<"EOT"
<your script here>
EOT
If you need some expansions to occur but want to disable others, then escape the ones that should not be expanded by putting a \ in front of them like you have done in your existing script.
Thanks to the feedback posted in the comments I was able to fix the issue. Posting a "fixed" version of the wrapper script below.
In short, the solution is to escape $SLURM_ARRAY_TASK_ID.
#!/bin/bash
for argument in "$#"
do
key=$(echo $argument | cut -f 1 -d'=')
value=$(echo $argument | cut -f 2 -d'=')
case "$key" in
"job_name") job_name="$value" ;;
"cpus") cpus="$value" ;;
"memory") memory="$value" ;;
"time") time="$value" ;;
"array") array="$value" ;;
*)
esac
done
{ tee /dev/stderr | sbatch; } <<EOT
#!/bin/bash
#SBATCH --account=foobar
#SBATCH --cpus-per-task=${cpus:-1}
#SBATCH --mem-per-cpu=${memory:-1}GB
#SBATCH --time=${time:-00:01:00}
#SBATCH --array=${array:-1-2}
#SBATCH --job-name=${job_name:-Default_Job_Name}
if [ -z "\$SLURM_ARRAY_TASK_ID" ]
then
echo "The array ID \$SLURM_ARRAY_TASK_ID is empty"
else
echo "The array ID \$SLURM_ARRAY_TASK_ID is NOT empty"
fi
python foo.py -a \$SLURM_ARRAY_TASK_ID
EOT
cat slurm-123456789_1.out yields :
The array ID 1 is NOT empty
array number is : 1
Note: the { tee /dev/stderr | sbatch; } is not necessary, but is very useful for debugging (thanks Charles Duffy)
Related
I have created a bash script which takes 2 command line arguments. It works absolutely fine.
I want to go a step further, and want to show the types of arguments it takes. i.e. if we run my_script.bash --help it should tell the desired arguments in its order.
my_script.bash is as follows
#!/bin/bash
X="$1" ## Name
Y="$2" ## address
echo " Mr./Ms. $X lives in $Y ."
Since it takes two arguments name and address, I want to show these when this bash is executed without any inputs or number of inputs or using my_script.bash --help command.
ie executing ./my_script.bash or ./my_script.bash --help should show like below
$ ./my_script.bash
>>this script takes two arguments. please enter **Name** and **address**
Since these arguments are position specific so we cannot change the positions of Name and Address. It would be great if we could pass the arguments by defining --name --address.
$ ./my_script.bash --address Delhi --name Gupta
>> Mr./Ms. Gupta lives in Delhi .
Any help would be appreciated.
A basic option parsing loop uses while and case, and looks like this:
print_help ()
{
cat <<-EOF
${0##*/} - process name and address
--name NAME your name
--address ADDRESS your address
--help this help
EOF
}
die ()
{
echo "$#" >&2
exit 1
}
while [[ $# -gt 0 ]]; do
case $1 in
--help)
print_help
exit
;;
--name)
shift || die "$1: requires input"
name=$1
;;
--address)
shift || die "$1: requires input"
address=$1
;;
*)
die "$1: invalid argument"
esac
shift
done
The order of options doesn't matter.
You can test for [[ $1 == --help ]], but since your script has 2 required arguments, you could simply print the help whenever the number of arguments is not equal 2:
if (( $# != 2 ))
then
echo You have to provide 2 arguments 1>&2
exit 1
fi
I am very new to Bash scripting, can someone explain to me how the $# and $? work in the following code?
#!/bin/bash
ARGS=3 # Script requires 3 arguments.
E_BADARGS=85 # Wrong number of arguments passed to script.
if [ $# -ne "$ARGS" ]
then
echo "Usage: `basename $0` old-pattern new-pattern filename"
exit $E_BADARGS
fi
old_pattern=$1
new_pattern=$2
if [ -f "$3" ]
then
file_name=$3
else
echo "File \"$3\" does not exist."
exit $E_BADARGS
fi
exit $?
From Learn Bash in Y minutes:
# Builtin variables:
# There are some useful builtin variables, like
echo "Last program's return value: $?"
echo "Script's PID: $$"
echo "Number of arguments passed to script: $#"
echo "All arguments passed to script: $#"
echo "The script's name: $0"
echo "Script's arguments separated into different variables: $1 $2..."
From https://www.gnu.org/software/bash/manual/html_node/Special-Parameters.html
$# Expands to the number of positional parameters in decimal.
$? Expands to the exit status of the most recently executed foreground pipeline.
$# shows the number of the script's arguments
$? shows the last script's return value
about arguments: echo "ARG[$#]" before if and then execute the script like
script.sh 1
the ouput will be
ARG[1]
Usage: g old-pattern new-pattern filename
and so on
the ouput of $? could be also used on the command line:
#shell>ls
file1.txt g inpu nodes_list
#shell>echo $?
0
#shell>ls FileNameNotFound
ls: FileNameNotFound: No such file or directory
#shell> echo $?
1
In bash exist special variables... and i write you some of then.
$#- this is an special variable that content inside the number of command line (you can just count how many parameters were entered) you passed to the script. tis variable also represent the last command line but its better do this ${!#}
$?- this one is very special cause its represents is your script is fine this variable holds the exit status of the previosly command... its a littler confusing but it work perfectly... when you end you script you can positional this variable at the end and if she return 0 value you scrip is perfect is true, if she return 1 or others you must check out your lines.
I am learning bash. And I would like to make a function which wrap another function in a temporal script file and execute it with sudo -u command in sub-shell.
The problem I encountered is the generated script cannot find the wrapped function although it is exported in the wrap function.
I append test cords below. Someone who finds problems, please let me know. Thank you very much.
main.sh
source "./display.sh"
source "./sudo_wrap.sh"
display_func "load success"
sudo_wrap_func '' 'display_func' '3' '4 5'
output, display.sh, sudo_wrap.sh and generated temporal file are appended below,
output
display_func : load success
export -f display_func
30481: line 5: display_func: command not found
display.sh
function display_func() {
echo "display_func : $#"
}
sudo_wrap.sh
function sudo_wrap_func() {
local sudo_user="${1:-root}"
local function_name="${2:?'function_name is null string.'}"
shift 2
local func_augs=( "$#" )
local script
# *** script : header ***
script="#!/bin/bash\n"
script="${script}\n"
# *** script : making augments for function ***
script="${script}augs=("
for aug in "${func_augs[#]}"
do
if [[ "${aug}" =~ [[:blank:]] ]]; then
script=" ${script} \"${aug}\""
else
script=" ${script} ${aug}"
fi
done
script="${script})\n"
local tmp_script_file="${RANDOM}"
echo -e "${script}" >> "${tmp_script_file}"
# *** script : calling function with augments ***
echo -e "${function_name} \"\${augs[#]}\"\n" >> "${tmp_script_file}"
echo "export -f "${function_name}"" >&2
export -f "${function_name}"
sudo -u"${sudo_user}" bash "${tmp_script_file}"
rm "${tmp_script_file}"
}
temporally generated file (in this case, file name is 30481)
#!/bin/bash
augs=( 3 "4 5")
display_func "${augs[#]}"
As I said in a comment, the basic problem is that sudo cleans its environment (including both variables and functions) before running the command (/script) as another user. This can be overridden with sudo -E, but only if it's explicitly allowed in /etc/sudoers.
But the problem is not insoluble; you just have to include the definition of the function in the script, so it gets recreated in that environment. bash even has a convenient command, declare -f display_func, that prints the function definition in the appropriate form (and declare -p variable does the same for variables). So you can use those to add the appropriate definitions to the script.
Here's a script I wrote to do this. I made a few other changes vs. your script: I take -u username to specify a different user to run as (so you don't have to pass '' as the first argument if you don't want to specify a different user). I also added -f functionname and -v variablename to "export" additional function and variable definitions into the script (in case the main function depends on them). I also create the temp script file in /tmp, and change ownership if necessary so it'll be readable by the other user.
#!/bin/bash
me="$(basename "$0")"
usage() {
echo "Usage: $me [-u user] [-f otherfunction] [-v variablename] function [args...]" >&2
}
tmp_script_file=$(mktemp "/tmp/${me}.XXXXXXXXXXXX") || {
echo "Error creating temporary script file" >&2
exit 1
}
echo "#!/bin/bash" > "$tmp_script_file" # Not actually needed, since we'll run it with "bash"
# Parse the command options; "-u" gets stored for later, but "-f" and "-v" write
# the relevant declarations to the script file as we go.
sudo_user=""
while getopts u:f:v: OPT; do
case "$OPT" in
u)
sudo_user="$OPTARG" ;;
f)
declare -f "$OPTARG" >>"$tmp_script_file" || {
echo "Error saving definition of function $OPTARG" >&2
exit 1
} ;;
v)
declare -p "$OPTARG" >>"$tmp_script_file" || {
echo "Error saving definition of variable $OPTARG" >&2
exit 1
} ;;
?) usage; exit 1 ;;
esac
done
shift $(($OPTIND-1))
if (( $# == 0 )); then # No actual command specified
usage
exit 1
fi
# Write the main function itself into the script
declare -f "$1" >>"$tmp_script_file" || {
echo "Error saving definition of function $1" >&2
exit 1
}
# Then the command to run it, with arguments quoted/escaped as
# necessary.
printf "%q " "$#" >>"$tmp_script_file"
# the printf above won't write a newline, so add it by hand
echo >>"$tmp_script_file"
# If the script will run as someone other than root, change ownership of the
# script so the target user can read it
if [[ -n "$sudo_user" ]]; then
sudo chown "$sudo_user" "$tmp_script_file"
fi
# Now launch the script, suitably sudo'ed
sudo ${sudo_user:+ -u "$sudo_user"} bash "$tmp_script_file"
# Clean up
sudo rm "$tmp_script_file"
Here's an example of using it:
$ foo() { echo "foo_variable is '$foo_variable'"; }
$ bar() { echo "Running the function bar as $(whoami)"; echo "Arguments: $*"; foo; }
$ export -f foo bar # need to export these so the script can see them
$ export foo_variable='Whee!!!' # ditto
$ # Run the function directly first, so see what it does
$ bar 1 2 3
Running the function bar as gordon
Arguments: 1 2 3
foo_variable is 'Whee!!!'
$ # Now run it as another user with the wrapper script
$ ./sudo_wrap.sh -f foo -v foo_variable -u deenovo bar 1 2 3
Running the function bar as deenovo
Arguments: 1 2 3
foo_variable is 'Whee!!!'
Note that you could remove the need to export the functions and variables by either running the script with source or making it a function, but doing that would require changes to how $me is defined, the usage function, replacing all those exits with returns, and maybe some other things I haven't thought of.
* the wording of the question is terrible, sorry!
I have some bash functions I create
test() {echo "hello wold"}
test2() {echo "hello wold"}
Then in my .bashrc I source the file that has the above function . ~/my_bash_scripts/testFile
In the terminal I can run test and get hello world.
is there a way for me to add parent variable that holds all my functions together. For example personal test, personal test2.
Similar to every other gem out there, I downloaded a tweeter one. All it's methods are followed by the letter t, as in t status to write a status, instead of just status
You are asking about writing a command-line program. Just a simple one here:
#!/usr/bin/env bash
if [[ $# -eq 0 ]]; then
echo "no command specified"
exit
elif [[ $# -gt 1 ]]; then
echo "only one argument expected"
exit
fi
case "$1" in
test)
echo "hello, this is test1"
;;
test2)
echo "hello, this is test2"
;;
*)
echo "unknown command: $1"
;;
esac
Then save it and make it an executable by run chmod +x script.sh, and in your .bashrc file, add alias personal="/fullpath/to/the/script.sh".
This is just very basic and simple example using bash and of course you can use any language you like, e.g. Python, Ruby, Node e.t.c.
Use arguments to determine final outputs.
You can use "$#" for number of arguments.
For example,
if [ $# -ne 2 ]; then
# TODO: print usage
exit 1
fi
Above code exits if arguments not euqal to 2.
So below bash program
echo $#
with
thatscript foo bar baz quux
will output 4.
Finally you can combine words to determine what to put stdout.
If you want to flag some functions as your personal functions; no, there is no explicit way to do that, and essentially, all shell functions belong to yourself (although some may be defined by your distro maintainer or system administrator as system-wide defaults).
What you could do is collect the output from declare -F at the very top of your personal shell startup file; any function not in that list is your personal function.
SYSFNS=$(declare -F | awk '{ a[++i] = $3 }
END { for (n=1; n<=i; n++) printf "%s%s", (n>1? ":" : ""), a[n] }')
This generates a variable SYSFNS which contains a colon-separated list of system-declared functions.
With that defined, you can check out which functions are yours:
myfns () {
local fun
declare -F |
while read -r _ _ fun; do
case :$SYSFNS: in *:"$fun":*) continue;; esac
echo "$fun"
done
}
I've gone around and around on the quoting stuff on http://tldp.org for bash and googled until I am blue in the face. I've also tried every obvious quoting scheme for this issue, and yet nothing works.
The problem seems to be that a space inside of a quoted argument in the command run at the end of the script is being interpreted as a separator instead of as a quoted space.
Behold, here's my script (I know full well I'm a noob so comments on my style and/or uneccessary syntax is cool with me, I'll learn):
#!/bin/bash
date=`date`
args="$#"
MSEND_HOME=/home/patrol/Impact #Path to the Impact Directory
integrationName=Introscope #Name of the integration
persistEnabled=1 #1 for Yes, 0 for No
persist=""
bufDir=$MSEND_HOME/tmp/$integrationName #DO NOT CHANGE
cellName=linuxtest #Cell name to forward events to
loggingEnabled=1 #1 for Yes, 0 for No
logFile=$MSEND_HOME/log/$integrationName.$cellName.log
die () {
if [ $loggingEnabled -eq 1 ]
then
echo >>$logFile "$#"
fi
exit 1
}
[ "$#" -ge 1 ] || die "$date - At least 1 argument required, $# provided" "$#"
# This is where you would parse out your arguments and form the following
# slots as a minimum for sending an event.
class=$2
msg=\"$3\"
# Parse the first argument and assign the correct syntax
if [[ $1 == "INFORMATIONAL" ]]
then
severity=INFO
elif [[ $1 == "WARN" ]]
then
severity=WARNING
elif [[ $1 == "CRIT" ]]
then
severity=CRITICAL
else
severity=INFO
fi
#Additional slots can be set, parse them all in this variable;
#e.g., additionalSlots="slot1=value1;slot2=value2;slot3=\"value 3\""
additionalSlots=""
cmd="$MSEND_HOME/bin/msend"
cmd="$cmd -q"
cmd="$cmd -l $MSEND_HOME"
if [ $persistEnabled -eq 1 ]
then
cmd="$cmd -j $bufDir"
fi
cmd="$cmd -n $cellName"
cmd="$cmd -a $class"
cmd="$cmd -m $msg"
cmd="$cmd -r $severity"
if [ $additionalSlots ]
then
cmd="$cmd -b $additionalSlots"
fi
$cmd || die "$date - msend exited with error $? | Original arguments: $args | Command: $cmd"
#echo "msend exited with error $? | Original arguments: $args | Command: $cmd"
The script is executed like this:
./sendEvent.sh "CRIT" "EVENT" "Test Event"
The error I get from the msend executable is that the arguments are wrong, but I'm logging the command line in it's entirety to a file and when I run that logged command in the shell interactively, it works.
Here's the log output:
Tue Oct 4 20:31:29 CDT 2011 - msend exited with error 27 | Original arguments: CRIT EVENT Test Event | Command: /home/patrol/Impact/bin/msend -q -l /home/patrol/Impact -j /home/patrol/Impact/tmp/Introscope -n linuxtest -a EVENT -m "Test Event" -r CRITICAL
So if I paste /home/patrol/Impact/bin/msend -q -l /home/patrol/Impact -j /home/patrol/Impact/tmp/Introscope -n linuxtest -a EVENT -m "Test Event" -r CRITICAL and run it, it works.
If I run the script like ./sendEvent.sh "CRIT" "EVENT" "TestEvent" it works. But I need that argument to allow spaces.
I'm on the track that it's an $IFS issue or something... maybe a difference between the interactive shell and the script environment.
I'd appreciate any insight from smarter people than me!
tl;dr - My command doesn't work when run from within a script, but does when the logged command syntax is used in an interactive shell.
Short answer: see BashFAQ #50.
Long answer: When bash parses a line, it parses quote marks before doing variable substitution; as a result, when you put quotes inside a variable, they don't do what you'd expect. You're actually passing an argument list including '-m' '"Test' 'Event"' '-r' -- those double-quotes aren't around the arguments, they're in the arguments.
In this case, the best solution is to build the command in an array rather than a string. Also, get in the habbit of putting double-quotes around variables (e.g. filenames) when you use them, to prevent confusion if they contain spaces. With those changes (and a few other tweaks), here's my version of your script:
#!/bin/bash
date="$(date)" # Backquotes are confusing, use $() instead
args=("$#") # Save the args in an array rather than mushing them together in a string
MSEND_HOME=/home/patrol/Impact #Path to the Impact Directory
MSEND_HOME="$HOME/tmp" #Path to the Impact Directory
integrationName=Introscope #Name of the integration
persistEnabled=1 #1 for Yes, 0 for No
persist=""
bufDir="$MSEND_HOME/tmp/$integrationName" #DO NOT CHANGE
cellName=linuxtest #Cell name to forward events to
loggingEnabled=1 #1 for Yes, 0 for No
logFile="$MSEND_HOME/log/$integrationName.$cellName.log"
die () {
if [ $loggingEnabled -eq 1 ]
then
echo >>"$logFile" "$#"
fi
exit 1
}
[ "$#" -ge 1 ] || die "$date - At least 1 argument required, $# provided" "$#"
# This is where you would parse out your arguments and form the following
# slots as a minimum for sending an event.
class="$2" # Quotes not strictly needed here, but a good habbit
msg="$3"
# Parse the first argument and assign the correct syntax
if [[ "$1" == "INFORMATIONAL" ]]
then
severity=INFO
elif [[ "$1" == "WARN" ]]
then
severity=WARNING
elif [[ "$1" == "CRIT" ]]
then
severity=CRITICAL
else
severity=INFO
fi
#Additional slots can be set, parse them all in this array;
#e.g., additionalSlots="slot1=value1;slot2=value2;slot3=value 3" # Don't embed quotes
additionalSlots=""
cmd=("$MSEND_HOME/bin/msend") # Build the command as an array, not a string
cmd+=(-q) # Could equivalently use cmd=("${cmd[#]}" -q), but this is simpler
cmd+=(-l "$MSEND_HOME")
if [ $persistEnabled -eq 1 ]
then
cmd+=(-j "$bufDir")
fi
cmd+=(-n "$cellName")
cmd+=(-a "$class") # Possible bug: $2 and #3 aren't required, but they're getting added unconditionally
cmd+=(-m "$msg") # These should probably be conditional, like additionalSlots
cmd+=(-r "$severity")
if [ -n "$additionalSlots" ]
then
cmd+=(-b "$additionalSlots")
fi
"${cmd[#]}" || die "$date - msend exited with error $? | Original arguments:$(printf " %q" "${args[#]}") | Command:$(printf " %q" "${cmd[#]}")"
#echo "msend exited with error $? | Original arguments:$(printf " %q" "${args[#]}") | Command:$(printf " %q" "${cmd[#]}")"
I think the arg goes wrong with this assignment: cmd="$cmd -m $msg".
Change it to cmd="$cmd -m \"$msg\"".
Okay, I don't see the exact problem immediately, but I can tell you what it is; this hint should help.
Remember that the shell quoting mechanism only interprets a string once. As a result, if you're not careful, what you thought was "foo" "a" "b" is in fact "foo a b" -- that is, all one token, not three.
Run the script with bash -x which will show you at each step what the shell is actually seeing.