getent is available on Linux, Solaris, BSD.
Not HP-UX, don't know about AIX.
Like the p-tools (pgrep, pkill, ptree, pargs, pfiles, ...),
getent simplifies debugging and shell scripting.
Seems nicer than niscat/ypcat !
BTW in the passwd|group|hosts-change case there is one reason for not using getent:
name service cache (nscd etc.) can obscure things. E.g. if an add-user script is re-run within the passwd cache time.
However, in the "files" case, nscd can see changes faster than the cache time...
Here is what I get if I put that modificatin of yours
RUID=`ssh $i awk -F: '\$3=="'$UNIQUE'"' /etc/passwd`
[root@GPGLNX02 test]# ./useradd.sh
awk: ==
awk: ^ syntax error
I even run it manually and it gives me a syntax error and the value that I am looking for
`ssh HOSTXXX awk -F: '\$3==2902' /etc/passwd`
awk: ==2902
awk: ^ syntax error
Strange seems to work fine for me on both AIX and Linux:
$ ssh HOSTXX awk -F: '\$3==0' /etc/passwd
root:x:0:0:root:/root:/bin/bash
Could the host system be Solaris? If so you may need nawk
My hosts are all rhel 4 and 5
How about using echo instead of awk?
$ ssh HOSTXX echo -F: '\$3==0' /etc/passwd
-F: $3==0 /etc/passwd
The error happens in `backticks`:
$ echo `ssh HOSTXX echo -F: '\$3==0' /etc/passwd`
-F: ==2902 /etc/passwd
`expression` differs from $(expression) here:
$ echo $(ssh HOSTXX echo -F: '\$3==0' /etc/passwd)
-F: $3==2902 /etc/passwd
1 Like
Is it usual to manage users by number manually across mny machines? People usually care more about the login user name id.
1 Like
Looks like three backslashes if backticks are used then:
RUID=`ssh $i awk -F: '\\\$3=="'$UNIQUE'"' /etc/passwd`
or
RUID=$(ssh $i awk -F: '\$3=="'$UNIQUE'"' /etc/passwd)
1 Like
Just a dum question.....
What does it mean by the $ sign infront of that bracket ()
Is that another way of assign the value of ssh..... to RUID
RUID=$(ssh $i awk -F: '\$3=="'$UNIQUE'"' /etc/passwd)
Yes, most new shell versions support either $(command) or `command` command substitution styles. I feel the newer $() style is easier to read and looks much better when nested.
1 Like
Bang on
This works for me
RUID=$(ssh $i awk -F: '\$3=="'$UNIQUE'"' /etc/passwd)
One more thing is, the home directory still doesn't created remotely by the script
here is the line of useradd in my script.
Could some please have a second look
`ssh $i useradd -u "$UNIQUE" -d "$HOMEPATH" -p $(openssl passwd "$PASSWORD") -c "$ROLE" -m -k /etc/skel/ "$UNAME"`
$UNIQUE-uid
$HOMEPATH-/home/$UNAME
$UNAME-username
$PASSWORD-password in text
The following is the error when I run the script, says
home directory already exists, but I double check remote hosts, nothing exist prior to running script
[root@HOSTXX test]# ./useradd.sh
=====User ID is currently available on HOSTXY, ready to add new user=====
Warning: truncating password to 8 characters
Creating mailbox file: File exists
useradd: warning: the home directory already exists.
Not copying any file from skel directory into it.
==========User1 created==========
=====User ID is currently available on HOSTXZ ready to add new user=====
Warning: truncating password to 8 characters
==========User1 created==========
Yes, the $() is nicer, suggest the actual subshell, nests, and supports vi % end finder and similar GUI IDE tools. Just make your 'case' test entries not "xxx*)" as in old examples, but the perfectly legal and balanced "(xxx*)", so the % and such are not confused.
1 Like
Thank you for all your post
It really means alot to me and it gave me a good bash learning curve
One more thing is, the home directory still doesn't creates remotely by the script
here is the line of useradd in my script.
Could someone review and advise what I did wrong here?
`ssh $i useradd -u "$UNIQUE" -d "$HOMEPATH" -p $(openssl passwd "$PASSWORD") -c "$ROLE" -m -k /etc/skel/ "$UNAME"`