https://pinot.apache.org/ logo
Join Slack
Powered by
# troubleshooting
  • s

    srisudha

    05/23/2020, 1:36 AM
    Yes surely
  • s

    srisudha

    05/23/2020, 1:36 AM
    From this blog.. https://engineering.linkedin.com/blog/2019/auto-tuning-pinot
  • m

    Mayank

    05/23/2020, 1:36 AM
    Do you think your data as scope for aggregation on the fly?
  • m

    Mayank

    05/23/2020, 1:37 AM
    As in incoming events with duplicate dimensions?
  • s

    srisudha

    05/23/2020, 1:37 AM
    Yes v need that for future
  • m

    Mayank

    05/23/2020, 1:38 AM
    I’ll look at why auto tuning is recommending such tiny segments
  • m

    Mayank

    05/23/2020, 1:39 AM
    Cc @Neha Pawar ^^
  • s

    srisudha

    05/23/2020, 1:40 AM
    V will also re run the utility , could be that with our initial understanding, v did not provide the right data to utility..
  • s

    srisudha

    05/23/2020, 1:41 AM
    Data with different dimensions and v surely need queries that should aggregate over those different dimensions
  • d

    Dan Hill

    05/27/2020, 7:03 PM
    I'm working on a layer that joins MySQL and Pinot data that requires interactive latencies (<100ms). I've been trying to use Presto for this but I haven't found configurations that make this scale well (Presto is not designed for interactive latencies). I'll dive deeper into Presto soon to see where the issues are (e.g. are the connectors causing requests to be executed serially?). What other ways people join Pinot and other data sources?
  • k

    Kishore G

    05/27/2020, 7:04 PM
    query time joins are unpredictable and hard to achieve low latency
  • k

    Kishore G

    05/27/2020, 7:05 PM
    https://www.slideshare.net/KishoreGopalakrishna/history-of-apache-pinot/27?src=clipshare
  • k

    Kishore G

    05/27/2020, 7:06 PM
    is there possibility of pre-join?
  • d

    Dan Hill

    05/27/2020, 7:11 PM
    Like denormalize the data out of Pinto? Yes, it's an option. Not ideal since there would be a delay to the stats.
  • d

    Dan Hill

    05/27/2020, 7:12 PM
    Usually, I'll be joining like 100 rows from MySQL with Pinot aggregate stats.
  • d

    Dan Hill

    05/27/2020, 7:12 PM
    These are not big joins.
  • d

    Dan Hill

    05/27/2020, 7:12 PM
    Presto just has a bunch of overhead (even for small queries)
  • k

    Kishore G

    05/27/2020, 7:13 PM
    ah, is this more for decoration?
  • d

    Dan Hill

    05/27/2020, 7:18 PM
    I'm not sure I understand 'decoration'. It's more that most of the query result sizes will be small and I can prevent the views if they are large. I currently do a
    MySQL left join Pinot
    to add Pinot stats onto other entities.
  • e

    Elon

    05/29/2020, 8:24 PM
    What is the name of the function in pinot to get the current timestamp, i.e. is it
    now()
    or is there a magic constant like
    current_timestamp
    ?
  • k

    Kishore G

    05/29/2020, 8:26 PM
    now()
  • e

    Elon

    05/29/2020, 8:40 PM
    Thanks! Is it newer than 0.3.0?
  • k

    Kishore G

    05/29/2020, 8:40 PM
    yes
  • k

    Kishore G

    05/29/2020, 8:41 PM
    dont know if it will be part of 0.4.0
  • k

    Kishore G

    05/29/2020, 8:41 PM
    @Xiang Fu @Haibo Wang
  • x

    Xiang Fu

    05/29/2020, 8:41 PM
    it’s part of 0.4.0
  • e

    Elon

    05/29/2020, 8:41 PM
    Nice!
  • x

    Xiang Fu

    05/29/2020, 8:41 PM
    now()
  • e

    Elon

    05/29/2020, 8:55 PM
    Sorry, 1 more question, we found a workaround but need to convert a double to an int in pinot calcite sql, is there a way?
  • k

    Kishore G

    05/29/2020, 8:58 PM
    example
1...106107108...166Latest