Details
-
Bug
-
Resolution: Done
-
Critical
-
4.0.0
-
Security Level: Public
-
None
-
Sherlock 1891, CentOS6 64bit
-
Untriaged
-
-
Unknown
Description
Edit: Originally I thought I was creating GSI indexes with CREATE PRIMARY INDEX ..., turns out that creates mapReduce indexes by default. However there potentially an even worse problem here in that an idle indexer process is using more memory than the quota.
I created a three-node cluster, with nodes 1 and 2 just for KV, and node3 for index and query service. I then loaded the travel-sample bucket (in addition to default.
I issued a couple of queries (using mapReduce views):
SELECT * FROM `travel-sample` WHWwhere country == "United States";
|
SELECT COUNT(*) FROM `travel-sample` where country == "United States";
|
I then looked at the memory usage on node3 (index / query node) and was surprised to see:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
|
9008 couchbas 20 0 944m 448m 7688 S 0.0 11.7 0:07.20 indexer
|
448MB resident, 944MB VSZ.
Given there are no GSI indexes created, and I specified 256MB of Index quota, why is indexer using 448MB RAM?