[Linux-HA] How to prevent unexpected unmount (stop) of ordered clone Filesystem resource on alive node

[Takekazu Okamoto]

Andrew-san,

Thank you for your comment.

I tried modules of 2.1.3 on following location.
http://linux-ha.org/download/index.html#2.1.3
My system is built from scratch again.
Unfortunately I have faced same problem.

>> 2.1.3
Dec 29 20:47:23 node2 heartbeat: [3301]: info: Configuration validated.
Starting heartbeat 2.1.3
Dec 29 20:47:23 node2 heartbeat: [3336]: info: heartbeat: version
2.1.3
<<

In addition, following change was made.
>> Filesystem RA
node2:/usr/lib/ocf/resource.d/heartbeat # diff -u Filesystem.org
Filesystem
--- Filesystem.org      2007-12-29 20:17:38.000000000 +0900
+++ Filesystem  2007-12-29 20:18:06.000000000 +0900
@@ -477,7 +477,7 @@
                        # Duplicate removal - start can contain nodes
                        # already on the active list, confusing the
                        # script later on:
-                       for UNAME in "$n_active"; do
+                       for UNAME in $n_active; do
                                n_start=`echo ${n_start} | sed
s/$UNAME//`
                        done
                        # Merge pruned lists again:
@@ -490,7 +490,7 @@
                stop)
                        # remove unames from notify_stop_uname; these
have been
                        # stopped and can no longer be considered
active.
-                       for UNAME in "$n_stop"; do
+                       for UNAME in $n_stop; do
                                n_active=`echo ${n_active} | sed
s/$UNAME//`
                        done
                        ;;
<<

Because
>> error
Dec 29 20:13:10 node2 lrmd: [3805]: info: RA output:
(configstoreclone:1:start:stderr) sed: -e expression #1, char 0: no
previous regular expression
Dec 29 20:13:13 node2 lrmd: [3805]: info: RA output:
(configstoreclone:0:start:stderr) sed: -e expression #1, char 7:
unterminated `s' command
<<

Could you let me know about something wrong?

Thank you,
Takekazu Okamoto

>>> "Andrew Beekhof" <beekhof at gmail.com> 2007/12/29 3:51 >>>
In the version you have, the clone placement wasn't completely stable
(which is my term for the behavior you want :-)
This will be addressed in SP2 which is due out "soon" IIRC

If you can't wait for SP2, try either the latest interim build or
2.1.3