You can have also the case both nodes crash ... The bottom line is that a write loss can occur in any system. I am always surprised to hear even senior consultants saying that in a high reliability database no write loss can occur or the risk is low (think about the human factor! Eg an admin accidentally shuts down both or some automated maintenance procedure of the underlying operating system).
> On 13 Sep 2016, at 09:59, Christian Reiniger <christian.reini...@iisy.de> > wrote: > > Hi, > > I’ve been trying to find information on cache persistence write behind > behavior that might lead to entries not being written to the persistent > store. Most notably: what happens in the following scenario: > > · Cache ‘cache’ has two backup copies on nodes A and B respectively > and is configured with persistence to a RDBMS, with write behind activated. > · Node A collects some write operations, but crashes before the > entries can actually be written. > > Will the loss of node A cause node B jump in and write these entries or can > ignite “forget” to write them? > > Thank you > -- > > Christian Reiniger > >
