Open GoogleCodeExporter opened 9 years ago
Zach - maybe you can own this one? (Seems like LSM-as-secondary index, and
index lifecycle, know-how will be crucial.) Thx!
Original comment by dtab...@gmail.com
on 27 Jan 2014 at 8:47
Is this issue blocking you Pouria?
Also, would it be possible to get a more minimal reproducer? Mike's thoughts
(which I agree with) are that this is probably not related to parallelism so a
single node reproducer might be possible--including using a much smaller data
volume.
If it is not urgent then I'd prefer to wait until we complete the LSM
experiments before looking at this issue since the deadline is approaching fast.
Original comment by zheilb...@gmail.com
on 29 Jan 2014 at 6:46
Not a blocking issue at this point. As I mentioned in "Observation 2" in
issue's description, It is possible to execute the query successfully.
I will try to reproduce it on a smaller setup ...
Original comment by pouria.p...@gmail.com
on 29 Jan 2014 at 6:59
To be clear on my thought: You can only get us to do an index NL join if you
"hint" that we should. My guess is that we have VERY few test cases, so what's
happening could be something very, very basic that we simply have not tested
before. (This was fairly untrodden code until recently.)
Original comment by dtab...@gmail.com
on 29 Jan 2014 at 10:28
Original issue reported on code.google.com by
pouria.p...@gmail.com
on 24 Jan 2014 at 5:24