Marshall D. Lewis wrote:
> As I understand it, the recommended configuration requires at least (db
> size/# nodes) * 2 of ram per node. For example (using our numbers)
> (36G/3) * 2 = 24G
# of nodes must be a power of NoOfReplicas, which defaults to 2, so you
can't have 3 nodes. 4, 8, 16, etc... (I recall reading that it was
intended to be a multiple of NoOfReplicas, not a power, but in my own
testing, the cluster has only been stable when using 2,4,8 nodes)
> Is that correct? If only have 8G of ram on those servers would we have
> to give them 16G+ of swap in order to run a cluster? What effect would
> that have on performance...
Mikael Ronstrom posted this about a week ago
>>Not really recommendable to use less memory than the virtual memory
>>allocated by the ndbd process. We don't test such scenarios.
>>Probably it works with good performance most of the time, then some
>>time with lousy performance (due to swapping stopping the process for
>>every page-in and this will cause problems with the heartbeat
>>mechanisms.
>>
>>Particularly a full table scan over the table would be nasty since it
>>will touch the entire tables memory pages and will do so one at a
>>time.
> In one spot in the documentation it lists column types for ndb tables
> and text and blob types are not in the list. Does that mean that we
> would not be able to put table using those types into the cluster?
Again, Mikael posted this about 3 weeks ago
>>TEXT/BLOB attributes is very soon completed and pushed to the 4.1
>>clone. Fulltext Indexes on TEXT/BLOB attributes is only supported on
>>MyISAM and no short-term plans on doing those in MySQL Cluster.
Hope all that helps!
Devananda
Neopets, Inc
