Indexer MutationCount increases but no update queries sent


#1

The indexer process seems stuck at 100% using all 4 cores of one of 3 nodes in my cluster.

It has been running at high cpu for about 24 hours. This morning I thought I should increase the dedicated memory to the index service from 1GB to 2GB, but doesn’t seem to help.

When looking at the indexer logs, I notice entries like this:

2018-02-12T16:38:25.706+00:00 [Info] logReaderStat:: MAINT_STREAM MutationCount 21850000

the MutationCount value increases a lot, from 28500000 to 28600000 in just 2 seconds. But I’m not sending “any” update query to the cluster, I’m not sending any queries at all (just have the couchbase dashboard open)

I use only one bucket, 3,000,000 documents, 29 indexes (each for a single field).

All 3 nodes run Community Edition 5.0.1 build 5003 , the index path is a separate ssd drive than the data drive.

My main concern is the high cpu usage, let me know if you would need more info.

Thanks


#2

of course, I sent the email and then the cpu usage starting going down, first to just one core at 100% and now only about 4%, so, it seems that increasing the memory for the index service helped it catch up with whatever it was doing before.