Hi Neil et al,
I recommend using $(...) instead of `...` when making a change to that
line anyway.
Also note that if ${LOCALNODENAME} is empty (and that could happen if
scha_cluster_get dunps core), then you would get all entries of the host
database. If that is something that would confuse the code later, then I
recommend to check for non-empty ${LOCALNODENAME} before proceeding.
Thought I mention this, since I saw that happening with a different
agent just recently.
Otherwise change looks good to me.
Greets
Thorsten
Neil Garthwaite wrote:
> Hi,
>
> Please could you review this small change for HA-DHCP.
>
> http://cr.opensolaris.org/~neilg/colorado-dhcp-6831051
>
> The problem description is as follows:
>
> When HA-DHCP fails over the DHCP client table from one node to another
> it runs pntadm (DHCP network table management utility) in batch mode
> using a generated file that contains the new server IP, i.e.
>
> # cat /tmp/dhcp_dhcp-server-rs.tmp
> pntadm -M 10.11.159.163 -s 10.11.159.47
> pntadm -M 10.11.159.162 -s 10.11.159.47
>
> However, if duplicate entries exist within /etc/hosts then the batch
> file is built incorrectly, i.e.
>
> cat /tmp/dhcp_dhcp-server-rs.tmp
> pntadm -M 10.11.159.163 -s 10.11.159.47
> 10.11.159.47 10.11.159.0 <-- wrong
> pntadm -M 10.11.159.162 -s 10.11.159.47
> 10.11.159.47 10.11.159.0 <-- wrong
>
> Consequently, line 623 "${PNTADM} -B ${TMPFILE}" within functions.ksh
> fails with,
>
> pntadm [-r (resource)] [-p (path)] [-u (uninterpreted data)] (options)
> [(network ip or name)]
>
> The fix ensures that only unique entries are returned from a
> "/bin/getent hosts" call.
>
> Regards
> Neil
--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Sitz der Gesellschaft:
Sun Microsystems GmbH, Sonnenallee 1, D-85551 Kirchheim-Heimstetten
Amtsgericht Muenchen: HRB 161028
Geschaeftsfuehrer: Thomas Schroeder, Wolfgang Engels, Dr. Roland Boemer
Vorsitzender des Aufsichtsrates: Martin Haering
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
