FAQ
We fixed a lot of issues in the 1.1.1 release but without more information,
it's hard to be sure.

Are you able to repro this every time? If you just run that query
repeatedly, do you hit this?
The Impala daemon has a memory info page on the debug webpage (be default,
<host>:25000/memz).
Do you see the memory tracked on that debug page go up after each query?

Thanks
Nong

On Mon, Oct 21, 2013 at 1:46 AM, Klausen Schaefersinho wrote:

Hi,

we have an impala (Server version: impalad version 1.0.1 RELEASE (build
df844fb967cec8740f08dfb8b21962bc053527ef)) cluster with several nodes
running, and usually everything works quite fine. However we have
experienced that after some time certain queries (that require group by
etc) stopped working. Here is an example for such a query and the shell
output:

Query: select c.counts, count(c.counts) FROM ( SELECT count(*) AS counts
FROM table GROUP BY id) c GROUP BY c.counts
Query finished, fetching results ...
ERROR: Memory limit exceeded
ERROR: Invalid query handle

This looks to me pretty much like a memory leak. We have looked at the
logs but didn't find anything specific. After restarting the server the
query worked again.

Is this a know issue? Does the latest version of Impala fix this?


THX,

Klaus Schaefers

To unsubscribe from this group and stop receiving emails from it, send an
email to impala-user+unsubscribe@cloudera.org.
To unsubscribe from this group and stop receiving emails from it, send an email to impala-user+unsubscribe@cloudera.org.

Search Discussions

  • Klausen Schaefersinho at Oct 21, 2013 at 5:27 pm
    Hi Nong,

    Until now I could reproduce the issue. I will try to write a little script
    that fires the query the entire night... Lets see what happens.


    Cheers,

    Klaus

    On Mon, Oct 21, 2013 at 6:54 PM, Nong Li wrote:

    We fixed a lot of issues in the 1.1.1 release but without more
    information, it's hard to be sure.

    Are you able to repro this every time? If you just run that query
    repeatedly, do you hit this?
    The Impala daemon has a memory info page on the debug webpage (be default,
    <host>:25000/memz).
    Do you see the memory tracked on that debug page go up after each query?

    Thanks
    Nong


    On Mon, Oct 21, 2013 at 1:46 AM, Klausen Schaefersinho <
    klaus.schaefers@gmail.com> wrote:
    Hi,

    we have an impala (Server version: impalad version 1.0.1 RELEASE (build
    df844fb967cec8740f08dfb8b21962bc053527ef)) cluster with several nodes
    running, and usually everything works quite fine. However we have
    experienced that after some time certain queries (that require group by
    etc) stopped working. Here is an example for such a query and the shell
    output:

    Query: select c.counts, count(c.counts) FROM ( SELECT count(*) AS counts
    FROM table GROUP BY id) c GROUP BY c.counts
    Query finished, fetching results ...
    ERROR: Memory limit exceeded
    ERROR: Invalid query handle

    This looks to me pretty much like a memory leak. We have looked at the
    logs but didn't find anything specific. After restarting the server the
    query worked again.

    Is this a know issue? Does the latest version of Impala fix this?


    THX,

    Klaus Schaefers

    To unsubscribe from this group and stop receiving emails from it, send an
    email to impala-user+unsubscribe@cloudera.org.
    To unsubscribe from this group and stop receiving emails from it, send an
    email to impala-user+unsubscribe@cloudera.org.
    To unsubscribe from this group and stop receiving emails from it, send an email to impala-user+unsubscribe@cloudera.org.

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
groupimpala-user @
categorieshadoop
postedOct 21, '13 at 4:54p
activeOct 21, '13 at 5:27p
posts2
users2
websitecloudera.com
irc#hadoop

2 users in discussion

Klausen Schaefersinho: 1 post Nong Li: 1 post

People

Translate

site design / logo © 2022 Grokbase