On 11/6/2010 8:26 PM, Jonathan Ellis wrote: > On Sat, Nov 6, 2010 at 4:51 PM, Reverend Chip <rev.c...@gmail.com> wrote: >> Am I to understand that >> ring maintenance requests can just fail when partially complete, in the >> same manner as a regular insert might fail, perhaps due to inter-node >> RPC overflow? > Yes, in beta3 this can happen. This was fixed in CASSANDRA-1676.
Interesting. I'll upgrade. Meanwhile, I waited until the next day, (!) and the cluster reformed somewhere in the meantime. I'm now seeing streaming to the new node. Is there an existing tool to just read everything from every node, just to force a read repair on everything? >> It would appear, then, that Cassandra isn't designed to be operated and >> understood without constant log watching of all nodes. > Not in beta, it's not. :) > > (In fact I would recommend running beta nodes at debug log level so > when something goes wrong you have a better picture of what happened.) Point taken :-)
