Benjamin Arntzen

I think it's worth saying... Over and over if necessary... DFS is hard to do well, and we should be so lucky to have multiple good projects. MooseFS is there...

Huge +1 especially if it can be ported to Red Hat OpenShift Origin

I'm in the process of setting up a 10gbit lab in a similar setup to you. I'll try to report back with some results later.

Yep, in most public clouds you cannot use the uRaft solution. Or even other ARP-based solutions such as Pacemaker, keepalived etc. There should be some ways around it - some...

If you can figure out the rest of the glue (starting/stopping/moving around masters etc etc) here's an example of doing something like this in Pacemaker/Corosync - https://medium.com/@gt.anand1994/ha-cluster-with-elasticip-using-corosync-and-pacemaker-a013d288ae8

I believe that's roughly how MooseFS PRO works.

@eleaner no, in the PRO version of MooseFS I believe the clients know about all masters. It uses DNS to discover them, but does not rely on failing until it...

I need stop/restart in my scripts and LSB demands it of init scripts.

@onlyjob maybe submit a pull request with new init scripts? On Dec 29, 2014 9:05 AM, "Marcin Sulikowski" [email protected] wrote: > We are aware of the fact that our init...

Hi @amokhuginnsson - I think you may have misinterpreted the other issue (sharding). It proposes a solution to split brain and other issues - master A is fully responsible for...