question on bash efficiency (when/howto source a function and execute it)

Question

I want to execute a this function on login as the postgres user

postgres@servcer:~/.bash.d/functions> cat check_pg-instances.func

#!/usr/bin/bash

# ##############################################################################
#
# function to detect all running postgres instances and check them with pg_isready
#
# ##############################################################################

is_primary()    {
                pg_port=$1
                primary=$(  psql -p "${pg_port}" -tc "SELECT state FROM pg_stat_replication ;" )
                secondary=$(psql -p "${pg_port}" -tc "SELECT client_hostname FROM pg_stat_replication ;" \
                        | cut -d '.' -f1 | sed "s/^ //" )
                }


is_replica()    {
                pg_port=$1
                replica=$(psql -p "${pg_port}" -tc "SHOW primary_conninfo ;" | head -1 | sed "s/^ //" )
                main=$(   psql -p "${pg_port}" -tc "SHOW primary_conninfo ;" | head -1 | sed "s/^ //" \
                        | awk {'print $3'} | sed "s/'//g" | cut -d '=' -f 2 | cut -d '.' -f 1 )
                }


replication_status()    {
                        if [[ -n ${primary} ]] ; then
                                echo replicated by \> "${secondary}"
                        elif [[ -n ${replica} ]] ; then
                                echo replicating from \< "${main}"
                        else
                                echo STANDALONE
                        fi
                        }

function check_pg-instances() {

      # make sure we are at the right location #################################
      cd "$PGBIN" || exit

      # variables ##############################################################
      local instance_names=($(grep PGPORT= .pg*.env | sort -t '=' -k2,2 | cut -d '.' -f2  |  sed "s/^pg-//" ))
      local ports=($(grep PGPORT= .pg*.env | cut -d '=' -f2 | sort -t '=' -k2,2 | tr '\n' ' ' ))
      local counter=$( echo "${instance_names[@]}" | wc -w )
      local count=0

      echo ; echo ================================================================================

      while [[ "${counter}" -gt 0 ]] ; do


         if $(psql -p ${ports[$count]} -l >/dev/null 2>&1) ; then
             is_primary ${ports[$count]}
             is_replica ${ports[$count]}
         fi

         echo -e "postgres $( pg_isready -p ${ports[$count]} | sed "s#/tmp:${ports[$count]} - ##" ) | port: ${ports[$count]} | ${instance_names[$count]} \
         | $(replication_status)"

         counter=$(( "${counter}" - 1 ))
         count=$(( "${count}" + 1 ))
      done

      echo ================================================================================ ; echo

}

currently I handle this in ~/.bashrc like

# User specific functions ######################################################
test -d ~/.bash.d/functions && . ~/.bash.d/functions/*.func || true

# check postgres instances #####################################################
check_pg-instances

Now my question is whether this is clever or maybe better to put all this (loading the functions/*.func files and executing check_pg_instances) into a script and execute that on startup.

I could imagine in the way I am doing it right now I might have that function(s) in memory during the entire session and not only when it is actually needed.

Answer 1

Yes, you will have all 2 kB of these functions in memory.

Worrying about whether bash uses 2 kB of RAM, i.e., less than a memory page, on a database server, is very silly.

Declare these functions wherever convenient.

As usual for any performance considerations, the null hypothesis is "it doesn't matter". You can assume that safely, unless it uses a considerable amount of your overall RAM, or if it happens many thousand times per second. These shell functions are orders of magnitude below either criterion, so they don't matter.