Status: New
Owner: ----
New issue 939 by [email protected]: Maintain node group diversity during
auto-promotion
https://code.google.com/p/ganeti/issues/detail?id=939
What software version are you running? Please provide the output of "gnt-
cluster --version", "gnt-cluster version", and "hspace --version".
$ gnt-cluster --version
gnt-cluster (ganeti v2.10.1) 2.10.1
$ hspace --version
hspace (ganeti) version v2.10.1
compiled with ghc 7.4
running on linux x86_64
What distribution are you using?
Ubuntu Precise
What steps will reproduce the problem?
Setup: Ganeti cluster with 60 nodes split into 3 node groups (20 nodes
each). All nodes are master_capable. In this setup each node group is a
failure domain (physical rack of machines).
While performing a maintenance that requires all nodes to be evacuated,
marked offline, and rebooted I noticed that auto-promote was not keeping an
even distribution of master candidates across the node groups. I even
experienced the first node group loose all but one master candidate.
What is the expected output? What do you see instead?
I would expect Ganeti to *try* to distribute the master candidates across
the node groups if possible. I don't think this will always be possible
since users may explicitly prevent nodes from becoming a master candidate
using the master_capable flag.
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings
