High Number of Operations Possibly Killing Write?


(Praetor) #1

Hello - Can someone help me understand what the OPERATIONS metrics below signify? we are facing read timeouts whilst POSTING to this specific index. We noticed a lot of Flush/Refresh/Merges (see below) on this index! Could they be the cause of the timeouts? also - its been like this for a few hours - any suggestions on how to "Cool" down this index to make it serviceable for writes?

INDEX: myapp-2018.05.12-000011

Documents
Primaries Total
Size 828.2 MB 1.6 GB
Documents 144,761 289,522
Deleted 0 0
Field Data Cache 0 B 0 B

Operations
Flush 30
Refresh 104,217
Merges 11,243


(Christian Dahlqvist) #2

Are you by any chance performing a lot of reasonably frequent updates? Which version are you on?


(Praetor) #3

yes - but its a one time bulk Write to this specific index. Version 5.5 - am curious if the writes die down - would this operations settle down?


(Christian Dahlqvist) #4

How frequently are you updating a single document? What is your average document size?


(Praetor) #5

we never update a document. its only inserts and its only a few KB in an extreme case. But we did do a large number inserts.
FYI I did check this morning and the stats haven't changed. The index is NOT in use in any read/write capacity for now. Are these stats a false negative?


(system) #6

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.