[Linux-cluster] Restarting cluster node, without other nodes hanging

Nikola Savic niks at logik-internet.rs
Sat Feb 26 02:41:45 UTC 2011


  As mentioned in the previous post, I have setup my first cluster with
3 nodes running CentOS 5.4. Two nodes are mirrored using DRBD, and that
is used as shared storage for cluster. Third node is accessing that
shared storage using GNBD. When all three nodes are up and running,
everything works well. However, when I need to shutdown one of nodes, I
usually can't do it without hanging one or all of nodes, and in most
cases I have to use hardware reset :( I tried stopping all services
using shared storage (apache web server), and then stopping all cluster
related services (cman, clvmd, rgmanager, gnbd_imports, etc.). For some
reason, it doesn't work.

  Can anyone tell me what steps should I take when one of the cluster
nodes needs to be restarted?



More information about the Linux-cluster mailing list