Metadata warning seems incorrect

Farshid Ghods added a comment - 04/Dec/12 1:40 PM MB-7128 is to address the fix for 2.0.1

Perry Krug added a comment - 05/Dec/12 4:12 AM Sorry, what does 7128 have to do with this bug?

Aleksey Kondratenko added a comment - 05/Dec/12 11:13 AM ep_max_checkpoints 2 ep_max_data_size 2097152000 ep_max_item_size 20971520 ep_max_size 2097152000 ep_max_txn_size 10000 ep_max_vbuckets 1024 ep_mem_high_wat 1572864000 ep_mem_high_wat 1572864000 ep_mem_low_wat 1258291200 ep_mem_low_wat 1258291200 ep_mem_tracker_enabled true ep_min_data_age 0 ep_mlog_compactor_runs 0 ep_mutation_mem_threshold 0 ep_num_access_scanner_runs 0 ep_num_eject_failures 1458007050 ep_num_expiry_pager_runs 1 ep_num_non_resident 2773442 ep_num_not_my_vbuckets 0 ep_num_ops_del_meta 0 ep_num_ops_get_meta 0 ep_num_ops_set_meta 0 ep_num_pager_runs 527 ep_num_value_ejects 2773442 ep_oom_errors 0 ep_overhead 4385112450 ep_pager_active_vb_pcnt 40

Aleksey Kondratenko added a comment - 05/Dec/12 11:13 AM Indeed ep_overhead is above max data size

Perry Krug added a comment - 05/Dec/12 11:23 AM And what do we measure for the "metadata in RAM" stat in the UI? It's the discrepancy there that I'm confused about...

Chiyoung Seo added a comment - 10/Dec/12 3:45 PM Mike, I think you already had the bug opened for this meta data issue. Please mark it as duplicate if so. We should fix this issue for 2.0.1 release.

Mike Wiederhold added a comment - 17/Dec/12 1:41 PM Also see: http://www.couchbase.com/issues/browse/MB-7218

Aleksey Kondratenko added a comment - 18/Dec/12 11:28 AM - edited Copying Mike's helpful explanation from hidden ticket: Alk, I have looked at this issue and similar issues and their are two issues here and one bug that I need you to fix on the ns_server side. The first issue here is that some of the XXX-use-case servers are heavily over loaded. In some of the logs I have looked at I saw over a 500k items in a lot of the checkpoint queues. As a result the server memory is overloaded and waiting to persist items to disk. The stat that is used by ns_server for the alerts is also no the correct stat. ep_overhead is use to keep track of the size of the checkpoint queues and other structures used by ep-engine. The correct stat to use (vb_active_meta_data + vb_replica_meta_data + vb_pending_meta_data). This should resolve the issue. Also, note that the issue of heavy checkpointing overhead will benefit greatly from having multiple writers in our next release.

Aleksey Kondratenko added a comment - 18/Dec/12 4:34 PM BTW, not having to parse 3 stats and add them together could help us (ns_server team). We have certain inefficiency parsing/handling larger count of stats.

Mike Wiederhold added a comment - 18/Dec/12 5:06 PM I will add this and assign the bug back to you once I'm done.

Mike Wiederhold added a comment - 28/Dec/12 8:31 PM http://review.couchbase.org/#/c/23618/ http://review.couchbase.org/#/c/23619/

Thuan Nguyen added a comment - 07/Jan/13 4:07 PM Integrated in github-ep-engine-2-0 #467 (See [ http://qa.hq.northscale.net/job/github-ep-engine-2-0/467/ ])      MB-7354 : Provide an aggregated meta data stat (Revision 5fab317edee5ce4af28f8f47ddb352cb3217a15e)      Result = SUCCESS Mike Wiederhold : Files : * docs/stats.org * src/ep_engine.cc