How to get max length of each column in unix?

This can be a general way to do it, so that you don't have to care about the number of fields you have: store the lengths in an array and keep checking if it is the maximum or not. Finally, loop through them and print the results.

awk -F'|' 'NR>1{for (i=1; i<=NF; i++) max[i]=(length($i)>max[i]?length($i):max[i])}           END {for (i=1; i<=NF; i++) printf "%d%s", max[i], (i==NF?RS:FS)}' file

See output:

$ awk -F'|' 'NR>1{for (i=1; i<=NF; i++) max[i]=(length($i)>max[i]?length($i):max[i])} END {for (i=1; i<=NF; i++) printf "%d%s", max[i], (i==NF?RS:FS)}' a2|4|6

For variable number of columns, we can store the maximum amount of columns in for example cols:

$ awk -F'|' 'NR>1{cols=(cols<=NF?NF:cols); for (i=1; i<=NF; i++) max[i]=(length($i)>max[i]?length($i):max[i])} END {for (i=1; i<=cols; i++) printf "%d%s", max[i], (i==cols?RS:FS)}' a2|4|6

This might work for you (but if there are many fields I'd use for cycle and an array to store the length of fields...):

awk -F '|' 'NR>1 {if ( length($1) > l1 ) { l1=length($1) }                  if ( length($2) > l2 ) { l2=length($2) }                  if ( length($3) > l2 ) { l3=length($3) }                 }             END { print l1 "|" l2 "|" l3 }' INPUTFILE