Error with ---ignore_fail_on_start=1 about mysql-master-ha HOT 5 OPEN

Use --ignore_fail_on_start , then it should work.
Document was incorrect, so I fixed the doc.

thanks!

the same situation and same conf
us1, us4 - alive. one of them is master
us3 - dead(slave)

i try change master manually

masterha_master_switch --master_state=alive --orig_master_is_new_slave 
--conf=/home/mha4mysql/etc/app1.cnf --new_master_host=us1

Tue Dec 18 14:41:07 2012 - [info] MHA::MasterRotate version 0.55.
Tue Dec 18 14:41:07 2012 - [info] Starting online master switch..
Tue Dec 18 14:41:07 2012 - [info]
Tue Dec 18 14:41:07 2012 - [info] * Phase 1: Configuration Check Phase..
Tue Dec 18 14:41:07 2012 - [info]
Tue Dec 18 14:41:07 2012 - [warning] Global configuration file 
/etc/masterha_default.cnf not found. Skipping.
Tue Dec 18 14:41:07 2012 - [info] Reading application default configurations 
from /home/mha4mysql/etc/app1.cnf..
Tue Dec 18 14:41:07 2012 - [info] Reading server configurations from 
/home/mha4mysql/etc/app1.cnf..
Tue Dec 18 14:41:07 2012 - [error][/usr/share/perl/5.14/MHA/MasterRotate.pm, 
ln93] Switching master should not be started if one or more servers is down.
Tue Dec 18 14:41:07 2012 - [info] Dead Servers:
Tue Dec 18 14:41:07 2012 - [info]   us3(172.16.50.13:3306)
Tue Dec 18 14:41:07 2012 - [error][/usr/share/perl/5.14/MHA/ManagerUtil.pm, 
ln178] Got ERROR:  at /usr/bin/masterha_master_switch line 53

could you add same option ignore_fail_on_start to masterha_master_switch? it'd 
be usefull i think. 
Sometimes it's needed to change master by some reason even if only 1 slave is 
dead.
thanks in advance!

--master_state=alive is intended to be invoked manually (for scheduled master 
switch) so if there is any dead slave I think you can easily remove from 
configuration file (by editing manually or running "masterha_conf_host 
--command=delete" command). As there are many workarounds, I'm not going to 
support the feature. (I'm open to accept patches:).

This seemed to allow me to ignore broken or lagging replication with the 
ignore_fail flag; I simply added this to 
/usr/lib/perl5/vendor_perl/MHA/ServerManager.pm

in sub check_replication_health {
...
    if ( $target->has_replication_problem($allow_delay_seconds) ) {
      $log->error(" failed!");
     ## FAF
      if( !$target->{ignore_fail} ) {
       croak;
      }