Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
>Hi Joseph, thanks for you answer.
>
>I'm using ocfs2 and o2cb driver cluster. With my settings, always when I have split brain situation one node is Standole and the other WFConnection.
>
>Only when I setup the after-sb-2pri disconnect, the two nodes are Standalone, so I execute the manual split brain recovery.
>
>I want setup a automatic policy to recovery.
Use text mail and reply on list to keep the thread complete and
benefit from the more experienced people on this list.
Anyway, I understood this from your initial post: I pointed you
to a doc that showed some possible configuration of drbd.conf
under similar circumstances.
If you read his config, he uses:
disk {
fencing resource-and-stonith;
}
Which needs a handler (He wrote his own):
handlers {
outdate-peer "/sbin/obliterate-peer.sh";
}
Note the net config difference? Both nodes are active, so this calls
possible for some different handling of SB symptom.
The difference is your using OCFS2 and not GFS{2} but it details a logical
procedure. Essentially, the config params you seem to be using are not
applicable to an active/active setup.
jlc