[jira] [Comment Edited] (PHOENIX-6671) Avoid ShortCirtuation Coprocessor Connection with HBase 2.x
[
https://issues.apache.org/jira/browse/PHOENIX-6671?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17510237#comment-17510237
]
chenglei edited comment on PHOENIX-6671 at 3/22/22, 12:12 PM:
--
[~larsh]，is HBASE-26869 fix the slow scan problem you mentioned? To me, I
think option 2 is more acceptable, because the {{User}} problem is hard to fix.
was (Author: comnetwork):
[~larsh]，whether HBASE-26869 fix the slow scan problem you mentioned? To me, I
think option 2 is more acceptable, because the {{User}} problem is hard to fix.
> Avoid ShortCirtuation Coprocessor Connection with HBase 2.x
> ---
>
> Key: PHOENIX-6671
> URL: https://issues.apache.org/jira/browse/PHOENIX-6671
> Project: Phoenix
> Issue Type: Bug
>Reporter: Lars Hofhansl
>Assignee: Lars Hofhansl
>Priority: Major
> Fix For: 5.2.0, 5.1.3
>
> Attachments: 6671-5.1.txt
>
>
> See PHOENIX-6501, PHOENIX-6458, and HBASE-26812.
> HBase's ShortCircuit Connection are fundamentally broken in HBase 2. We might
> be able to fix it there, but with all the work the RPC handlers perform now
> (closing scanning, resolving current user, etc), I doubt we'll get that 100%
> right. HBase 3 has removed this functionality.
> Even with HBase 2, which does not have the async protobuf code, I could
> hardly see any performance improvement from circumventing the RPC stack in
> case the target of a Get or Scan is local. Even in the most ideal conditions
> where everything is local, there was improvement outside of noise.
> I suggest we do not use ShortCircuited Connections in Phoenix 5+.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
[jira] [Comment Edited] (PHOENIX-6671) Avoid ShortCirtuation Coprocessor Connection with HBase 2.x
[ https://issues.apache.org/jira/browse/PHOENIX-6671?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17509356#comment-17509356 ] Kadir OZDEMIR edited comment on PHOENIX-6671 at 3/20/22, 12:05 AM: --- [~larsh], This failure is NOT related to your change. I run this test on my local machine and it failed without your patch. Let's file a bug on this. It is not a blocker for your patch. was (Author: kozdemir): [~larsh], This failure is related to your change. I run this test on my local machine and it failed without your patch. Let's file a bug on this. It is not a blocker for your patch. > Avoid ShortCirtuation Coprocessor Connection with HBase 2.x > --- > > Key: PHOENIX-6671 > URL: https://issues.apache.org/jira/browse/PHOENIX-6671 > Project: Phoenix > Issue Type: Bug >Reporter: Lars Hofhansl >Assignee: Lars Hofhansl >Priority: Major > Fix For: 5.2.0, 5.1.3 > > Attachments: 6671-5.1.txt > > > See PHOENIX-6501, PHOENIX-6458, and HBASE-26812. > HBase's ShortCircuit Connection are fundamentally broken in HBase 2. We might > be able to fix it there, but with all the work the RPC handlers perform now > (closing scanning, resolving current user, etc), I doubt we'll get that 100% > right. HBase 3 has removed this functionality. > Even with HBase 2, which does not have the async protobuf code, I could > hardly see any performance improvement from circumventing the RPC stack in > case the target of a Get or Scan is local. Even in the most ideal conditions > where everything is local, there was improvement outside of noise. > I suggest we do not use ShortCircuited Connections in Phoenix 5+. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Comment Edited] (PHOENIX-6671) Avoid ShortCirtuation Coprocessor Connection with HBase 2.x
[ https://issues.apache.org/jira/browse/PHOENIX-6671?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17509306#comment-17509306 ] Lars Hofhansl edited comment on PHOENIX-6671 at 3/19/22, 4:52 PM: -- We need to be careful with this in production environments too. The remote call will now always tie up an extra handler thread even when the target regions are local. On the other hand, the chance for that is p=1/number-of-regionservers, so unless there are special scenarios where we know the region is local I do not think there is any risk. Before we merge, let's see where HBASE-26812 is going. was (Author: lhofhansl): We need to be careful with this in production environments too. The remote call will now always tie up an extra handler thread even when the target regions are local. On the other hand, the chance for that is p=1/number-of-regionservers, so unless there are special scenarios where we know the region is local I do not think there is any risk. > Avoid ShortCirtuation Coprocessor Connection with HBase 2.x > --- > > Key: PHOENIX-6671 > URL: https://issues.apache.org/jira/browse/PHOENIX-6671 > Project: Phoenix > Issue Type: Bug >Reporter: Lars Hofhansl >Assignee: Lars Hofhansl >Priority: Major > Fix For: 5.2.0, 5.1.3 > > Attachments: 6671-5.1.txt > > > See PHOENIX-6501, PHOENIX-6458, and HBASE-26812. > HBase's ShortCircuit Connection are fundamentally broken in HBase 2. We might > be able to fix it there, but with all the work the RPC handlers perform now > (closing scanning, resolving current user, etc), I doubt we'll get that 100% > right. HBase 3 has removed this functionality. > Even with HBase 2, which does not have the async protobuf code, I could > hardly see any performance improvement from circumventing the RPC stack in > case the target of a Get or Scan is local. Even in the most ideal conditions > where everything is local, there was improvement outside of noise. > I suggest we do not use ShortCircuited Connections in Phoenix 5+. -- This message was sent by Atlassian Jira (v8.20.1#820001)
[jira] [Comment Edited] (PHOENIX-6671) Avoid ShortCirtuation Coprocessor Connection with HBase 2.x
[ https://issues.apache.org/jira/browse/PHOENIX-6671?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17509130#comment-17509130 ] Kadir OZDEMIR edited comment on PHOENIX-6671 at 3/19/22, 12:42 AM: --- [~larsh], the patch looks good to me. Would you please generate a PR just to trigger the pre-checkin tests (just to be sure this does not lead to some IT failures)? I think submitting a patch does not trigger these tests any more. was (Author: kozdemir): [~larsh], the patch looks good to me. Would you please you generate a PR just to trigger the pre-checkin tests (just to be sure this does not lead to some IT failures)? I think submitting a patch does not trigger these tests any more. > Avoid ShortCirtuation Coprocessor Connection with HBase 2.x > --- > > Key: PHOENIX-6671 > URL: https://issues.apache.org/jira/browse/PHOENIX-6671 > Project: Phoenix > Issue Type: Bug >Reporter: Lars Hofhansl >Priority: Major > Attachments: 6671-5.1.txt > > > See PHOENIX-6501, PHOENIX-6458, and HBASE-26812. > HBase's ShortCircuit Connection are fundamentally broken in HBase 2. We might > be able to fix it there, but with all the work the RPC handlers perform now > (closing scanning, resolving current user, etc), I doubt we'll get that 100% > right. HBase 3 has removed this functionality. > Even with HBase 2, which does not have the async protobuf code, I could > hardly see any performance improvement from circumventing the RPC stack in > case the target of a Get or Scan is local. Even in the most ideal conditions > where everything is local, there was improvement outside of noise. > I suggest we do not use ShortCircuited Connections in Phoenix 5+. -- This message was sent by Atlassian Jira (v8.20.1#820001)
