> when i use rfork (CLUSTERNODE_BEST) i get the error 'Remote Object'
> with strerror (errno).... and when i use rexecv (cmd, args,
> CLUSTERNODE_BEST) it picks the current node each time....
I just checked in a fix for the rfork error case. The strace fix
for rfork made CLUSTERNODE_BEST not work anymore.
As for the rexecve(), is the current node heavily loaded compared
to the other nodes? If the loads are about equal, it will always
choose the current node before considering going remote.
ps. i've attached a patchfile for the rfork fix
> also i noticed in the redhat branch the script cluster_start is issued
> at some point during the init stage, but it's not issued on debian...
> along with the noded daemon.
> CVIP.pm, ha-lvs the ha-lvs init script cannot find CVIP.pm, I've
> pushed the "/cluster/lib" path to the INC to fix it...
> i'm using the latest cvs update, i believe 1-0-0-RC5 or
> 1-STABLE-FOOBAR-FOO something.... and i've apt-get upgraded to the
> latest changes....