Fork me on GitHub
#crux
<
2020-12-12
>
bbss15:12:41

I think this is a short version of the query timeout bug I'm hitting, my guess this is an eql/project issue

(crux/submit-tx
 node
 [[:crux.tx/put
   {:crux.db/id :a
    :a/ref :b}]
  [:crux.tx/put
   {:crux.db/id :b}]])

(crux/q (crux/db node)
        '{:find [(eql/project ?a [* {:a/ref
                                     [*
                                      {:b/not-yet-added-attr [*]}]}])] ;;comment this attribute out for no timeout, works with in-mem db, not with kafka/rocksdb.
          :where
          [[?a :a/ref ?b]]
          :args [{?a :a}]})

refset11:12:36

Thanks again for trying to pin it down 🙂 it's definitely something to do with your doc-store configuration. I think what you're observing is that any queries which require hitting the doc-store via a local cache (i.e. anything using eql/project or large values that aren't fully covered by the index) are blocking indefinitely due to some kind of connectivity issue.

bbss11:12:53

Alright, it's happening on both the cloud machine and my local macOS machine, not sure if that matters. But note that the short example only includes two entities! So it's not a connectivity issue in the sense that it keeps going back and forth with high latency or so.

bbss11:12:33

it seems like if you try a join on an attribute that doesn't exist this happens.