stats in clustered mode
I'm viewing membase stats using the get_stats() method of the client API.
I noticed that when I have 2 server cluster (as opposed to 2 independent nodes), the "curr_items" count that each node reports is doubled. For example... after I set 10 items, each node would show a "curr_items" count of 20. Is this showing items on both nodes (including replicated items)?
Also, do you have any documentation on the ep_* stats and what they mean? Most of the stats provided are the same as regular memcached, which is documented. But the ep_* stats seem unique to membase/membase. Do any of these stats show us replication operations?
-Corey
Corey, check out [url]http://github.com/northscale/ep-engine/tree/master/docs[/url] for information on the ep-engine including descriptions of the stats.
For stats related to replication, you can use the 'stats tap' command. Realize that replication is built on top of TAP, so the stats are a bit lower level since TAP can do more than just replication.
Let me know if that helps Corey.
Thanks
Perry
These will also be available (both current and historical) via the REST API in beta 2, though it will be a (large) subset, possibly not every single numerical stat.
Corey, the doubling of "curr_items" is a known bug (I filed it just the other day) and will be included in an upcoming release.
I'll put together some documentation on the ep_* stats and get back to you.
Perry
Forum support is great for free but sometimes you need a guaranteed response time and dedicated resources for your questions or issues.
Consider purchasing enterprise-level support from Membase: http://www.membase.com/products-and-services/overview
Call or email "sales -at- membase -dot- com" today!