[Info-vax] reboot due to network failure

[Phillip Helbig---undress to reply]

In article
<2047d429-a530-4ecf-9c18-25efc98b2937 at c13g2000vbh.googlegroups.com>,
Volker Halle <volker_halle at hotmail.com> writes: 

> if nodes in a cluster loose connectivity for more than RECNXINTERVAL
> seconds, they remove all nodes, with which they can't communicate
> anymore, from their 'local view of the cluster'. 

Right.

In this case, due to the switch failure, all connectivity was probably 
lost at once.

> If communication is
> re-established, the 'sub-cluster' with the highest number of votes,
> nodes or SCSSYSTEMID will survive, all other nodes will crash with a
> CLUEXIT crash, automatically reboot and join the cluster again.

All the nodes have the same number of votes.  However, the one which 
survived does not have the highest SCSSYSTEMID.