On Wed, 2008-12-10 at 11:52 -0800, Konstantin Shvachko wrote: > This is probably related to HADOOP-4795.
Thanx for the observation and reference. However, my sense is that the bug report you reference reflects NameNode going into an infloop spin, whereas the situation we have faced concerns NameNode being stuck/hung, as though in a resource embrace error: Using no CPU, reacting to no outside events other than to accept incoming connections without proceeding to handle them. Can your referenced bug manifest itself as a NameNode hang as well as a spin?