Indexing large arrays

n1ql
index

#1

Hello -

We’re trying to create an index on properties inside a rather large array that can have up to 10,000 array items, and each item has the properties listed in the KEYS section of the index statement below.

The problem is the indexer runs for a while and then I think it eventually crashes because we see all the indexes on the “proddb06” node go into a “warmup” state then back to ready. Is it uncommon/untested to index arrays with 10,000 elements?

The server running the indexing service has 50% RAM free when the indexing starts and eventually makes it way up to 90% used before the query service appears to crash.

Here’s our create statement, note that “visitCharges” is the array that has up to 10,000 items per document.

CREATE INDEX idx_cm_charge_visitCharges
ON cloud_med(tenantName,DISTINCT ARRAY [vc.postDate,vc.cpt,vc.hcpcs,vc.chargeCode,vc.description] FOR vc IN visitCharges END)
WHERE type = “charges” AND tenantName = “someTenantName”
WITH {“nodes”:[“proddb06.8091”]}


#2

During CREATE INDEX statement query service crashes?
Check query.log, indexer.log and system messages any out of memory (/var/log/messages/…)