[Bigdata-developers] trivial ASK takes over 5s

[Jeremy J C. <jj...@sy...>]

As part of our health check that ensures we know when our system go down, we send the following query to all our blazegraph instances every minute:

ASK
FROM <http://does-not-exist.example.org/an/empty/named/graph>
FROM NAMED <http://does-not-exist.example.org/an/empty/named/graph>

WHERE {

}


Normally this is very quick!

However, during a period of heavy write activity using INSERT-by-POST commands that seem to be taking around 30s (this is from logs one level out, so there is room for doubt), we were seeing these queries occasionally take over 5s (the trivial ASK).

Is this expected?

My general belief was that writes and concurrent reads did not strongly conflict in blazegraph.


Jeremy