<?xml version="1.0" encoding="UTF-8"?>
<rss xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:taxo="http://purl.org/rss/1.0/modules/taxonomy/" version="2.0">
  <channel>
    <title>question Not Able to Trigger &amp;quot;Spill to Disk&amp;quot; On Impala, Documentation Extremely Unclear in Support Questions</title>
    <link>https://community.cloudera.com/t5/Support-Questions/Not-Able-to-Trigger-quot-Spill-to-Disk-quot-On-Impala/m-p/338709#M232906</link>
    <description>&lt;P&gt;I am trying to trigger spilling to disk on impala (Impala Shell v3.2.0-cdh6.3.4 )&amp;nbsp;&lt;BR /&gt;The cloudera documentation suggest the steps to do it here:&amp;nbsp;&lt;A href="https://docs.cloudera.com/documentation/enterprise/6/6.3/topics/impala_scalability.html" target="_blank" rel="noopener"&gt;https://docs.cloudera.com/documentation/enterprise/6/6.3/topics/impala_scalability.html&lt;/A&gt;&amp;nbsp;Under the section: "&lt;STRONG&gt;Testing performance implications of spilling to disk:&lt;/STRONG&gt;"&lt;/P&gt;&lt;P&gt;&amp;nbsp;&lt;/P&gt;&lt;P&gt;I am extremely confused by the two seemingly contradictory statements there: "&lt;SPAN&gt;Set the&amp;nbsp;&lt;/SPAN&gt;MEM_LIMIT&lt;SPAN&gt;&amp;nbsp;query option to a &lt;STRONG&gt;value that is smaller&lt;/STRONG&gt; than the peak memory usage reported in the profile output. &lt;STRONG&gt;Do not specify a memory limit lower than&lt;/STRONG&gt; reported in the profile output.&amp;nbsp;&lt;/SPAN&gt;"&lt;BR /&gt;So what should be the value of MEM_LIMIT ?&lt;BR /&gt;When I do set it to be lower than the peak memory usage, i get the following error : "Rejected query from pool root.centos: minimum memory reservation is greater than memory available to the query for buffer reservations. Memory reservation needed given the current...."&lt;BR /&gt;&lt;BR /&gt;When I set the&amp;nbsp;MEM_LIMIT to be higher, then I don't see any "SpilledPartitions" or "SpilledRuns"&lt;/P&gt;&lt;P&gt;counters, that is, not triggering any spilled to disk.&amp;nbsp;&lt;BR /&gt;&lt;BR /&gt;Since the documentation is unclear, I want to know how I can trigger spill to disk functionality without the queries failing ?&amp;nbsp;&amp;nbsp;&lt;/P&gt;</description>
    <pubDate>Tue, 15 Mar 2022 16:52:55 GMT</pubDate>
    <dc:creator>PratCloudDev</dc:creator>
    <dc:date>2022-03-15T16:52:55Z</dc:date>
    <item>
      <title>Not Able to Trigger "Spill to Disk" On Impala, Documentation Extremely Unclear</title>
      <link>https://community.cloudera.com/t5/Support-Questions/Not-Able-to-Trigger-quot-Spill-to-Disk-quot-On-Impala/m-p/338709#M232906</link>
      <description>&lt;P&gt;I am trying to trigger spilling to disk on impala (Impala Shell v3.2.0-cdh6.3.4 )&amp;nbsp;&lt;BR /&gt;The cloudera documentation suggest the steps to do it here:&amp;nbsp;&lt;A href="https://docs.cloudera.com/documentation/enterprise/6/6.3/topics/impala_scalability.html" target="_blank" rel="noopener"&gt;https://docs.cloudera.com/documentation/enterprise/6/6.3/topics/impala_scalability.html&lt;/A&gt;&amp;nbsp;Under the section: "&lt;STRONG&gt;Testing performance implications of spilling to disk:&lt;/STRONG&gt;"&lt;/P&gt;&lt;P&gt;&amp;nbsp;&lt;/P&gt;&lt;P&gt;I am extremely confused by the two seemingly contradictory statements there: "&lt;SPAN&gt;Set the&amp;nbsp;&lt;/SPAN&gt;MEM_LIMIT&lt;SPAN&gt;&amp;nbsp;query option to a &lt;STRONG&gt;value that is smaller&lt;/STRONG&gt; than the peak memory usage reported in the profile output. &lt;STRONG&gt;Do not specify a memory limit lower than&lt;/STRONG&gt; reported in the profile output.&amp;nbsp;&lt;/SPAN&gt;"&lt;BR /&gt;So what should be the value of MEM_LIMIT ?&lt;BR /&gt;When I do set it to be lower than the peak memory usage, i get the following error : "Rejected query from pool root.centos: minimum memory reservation is greater than memory available to the query for buffer reservations. Memory reservation needed given the current...."&lt;BR /&gt;&lt;BR /&gt;When I set the&amp;nbsp;MEM_LIMIT to be higher, then I don't see any "SpilledPartitions" or "SpilledRuns"&lt;/P&gt;&lt;P&gt;counters, that is, not triggering any spilled to disk.&amp;nbsp;&lt;BR /&gt;&lt;BR /&gt;Since the documentation is unclear, I want to know how I can trigger spill to disk functionality without the queries failing ?&amp;nbsp;&amp;nbsp;&lt;/P&gt;</description>
      <pubDate>Tue, 15 Mar 2022 16:52:55 GMT</pubDate>
      <guid>https://community.cloudera.com/t5/Support-Questions/Not-Able-to-Trigger-quot-Spill-to-Disk-quot-On-Impala/m-p/338709#M232906</guid>
      <dc:creator>PratCloudDev</dc:creator>
      <dc:date>2022-03-15T16:52:55Z</dc:date>
    </item>
    <item>
      <title>Re: Not Able to Trigger "Spill to Disk" On Impala, Documentation Extremely Unclear</title>
      <link>https://community.cloudera.com/t5/Support-Questions/Not-Able-to-Trigger-quot-Spill-to-Disk-quot-On-Impala/m-p/339385#M233076</link>
      <description>&lt;P&gt;Hi&amp;nbsp;&lt;a href="https://community.cloudera.com/t5/user/viewprofilepage/user-id/96578"&gt;@PratCloudDev&lt;/a&gt;&amp;nbsp;,&lt;/P&gt;&lt;P&gt;&amp;nbsp;&lt;/P&gt;&lt;P&gt;&lt;SPAN&gt;"Spill to disk" happens when there is no enough memory available&amp;nbsp;for a running query, Below is the example.&lt;/SPAN&gt;&lt;/P&gt;&lt;P&gt;&amp;nbsp;&lt;/P&gt;&lt;P&gt;&lt;SPAN&gt;Suppose you are running a query which is using 10gb ( per node peak memory) of memory and in case this query needs 12 gb of memory in this situation spill to disk happen on the configured scratch directories.&lt;/SPAN&gt;&lt;/P&gt;&lt;P&gt;&lt;SPAN&gt;you can see the directory by searching&amp;nbsp;the "Impala Daemon Scratch Directories" property in the impala configurations.&lt;/SPAN&gt;&lt;/P&gt;&lt;P&gt;&lt;SPAN&gt;If you do not want to fail the query then you need to make sure the configured scratch directories/disk has enough&amp;nbsp;space to store&amp;nbsp;spilling information, this can potentially be large amounts of data.&lt;/SPAN&gt;&lt;/P&gt;&lt;P&gt;&amp;nbsp;&lt;/P&gt;&lt;P&gt;&lt;SPAN&gt;Check the query profile for "per node peak memory" it is the actual memory used for that query on each daemon, suppose if it is 15GB then set the MEM_LIMIT to 10gb or 12gb to see the spill to disk functionality.&amp;nbsp;&lt;/SPAN&gt;&lt;/P&gt;&lt;P&gt;&lt;SPAN&gt;To understand why you are seeing the error[1] i need few details from your side.&lt;/SPAN&gt;&lt;/P&gt;&lt;P&gt;&amp;nbsp;&lt;/P&gt;&lt;P&gt;&lt;SPAN&gt;1. Screenshot of impala admission control pool settings.&lt;/SPAN&gt;&lt;/P&gt;&lt;P&gt;&lt;SPAN&gt;2. How much memory you are setting and seeing the below error[1]?&lt;/SPAN&gt;&lt;/P&gt;&lt;P&gt;&lt;SPAN&gt;3. Which pool you are using to run the query?&lt;/SPAN&gt;&lt;/P&gt;&lt;P&gt;&lt;SPAN&gt;4. If possible you can provide the query profile.&lt;/SPAN&gt;&lt;/P&gt;&lt;P&gt;&amp;nbsp;&lt;/P&gt;&lt;P&gt;&lt;SPAN&gt;Regards,&lt;/SPAN&gt;&lt;/P&gt;&lt;P&gt;&lt;SPAN&gt;Chethan YM&lt;/SPAN&gt;&lt;/P&gt;&lt;P&gt;&amp;nbsp;&lt;/P&gt;&lt;P&gt;&lt;SPAN&gt;[1].&amp;nbsp;Rejected query from pool root.centos: minimum memory reservation is greater than memory available to the query for buffer reservations.&lt;/SPAN&gt;&lt;/P&gt;</description>
      <pubDate>Wed, 23 Mar 2022 12:26:58 GMT</pubDate>
      <guid>https://community.cloudera.com/t5/Support-Questions/Not-Able-to-Trigger-quot-Spill-to-Disk-quot-On-Impala/m-p/339385#M233076</guid>
      <dc:creator>ChethanYM</dc:creator>
      <dc:date>2022-03-23T12:26:58Z</dc:date>
    </item>
  </channel>
</rss>

