[
https://issues.apache.org/jira/browse/IMPALA-13304?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Surya Hebbar updated IMPALA-13304:
----------------------------------
Description:
Currently, instance-level details of fragment events are completely omitted
from the experimental profile(V2) in contrast with the traditional profile.
This is to stop the profile size from blowing up with increasing number of
nodes.
Due to this, the query timeline's fragment timing diagram does not function.
The generated profiles can be switched by setting {{gen_experimental_profile}}
to {{true}} or {{{}false{}}}.
It would be beneficial to include aggregate instance-level metrics without
blowing up the profile size in case of large number of nodes. This would
further help in plotting the query timeline's fragment timing diagram as well.
was:
Currently, instance-level details of fragment events are completely omitted
from the experimental profile in contrast with the traditional profile. This is
to stop the profile size from blowing up with increasing number of nodes.
Due to this, the query timeline's fragment timing diagram does not function.
The generated profiles can be switched by setting {{gen_experimental_profile}}
to {{true}} or {{false}}.
It would be beneficial to include aggregate instance-level metrics without
blowing up the profile size in case of large number of nodes. This would
further help in plotting the query timeline's fragment timing diagram as well.
> Include aggregate instance-level metrics within experimental profile(V2)
> ------------------------------------------------------------------------
>
> Key: IMPALA-13304
> URL: https://issues.apache.org/jira/browse/IMPALA-13304
> Project: IMPALA
> Issue Type: Improvement
> Reporter: Surya Hebbar
> Assignee: Surya Hebbar
> Priority: Major
>
> Currently, instance-level details of fragment events are completely omitted
> from the experimental profile(V2) in contrast with the traditional profile.
> This is to stop the profile size from blowing up with increasing number of
> nodes.
> Due to this, the query timeline's fragment timing diagram does not function.
> The generated profiles can be switched by setting
> {{gen_experimental_profile}} to {{true}} or {{{}false{}}}.
> It would be beneficial to include aggregate instance-level metrics without
> blowing up the profile size in case of large number of nodes. This would
> further help in plotting the query timeline's fragment timing diagram as well.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
