[
https://issues.apache.org/jira/browse/YARN-5585?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15512673#comment-15512673
]
Varun Saxena commented on YARN-5585:
------------------------------------
Just to clarify, what I meant by having another index table was not to store
data in it. It only stores the entityID for
cluster!user!flow!run!app!entitytype and inverted created time.
The write to this table will only be when created time is reported i.e. when
application reports created time on start event (most probably).
As as part of the interface, we are claiming entities will be returned,
descendingly sorted by created time, I felt this use case we should definitely
support.
Whether we support sorting by some other parameter or not.
Currently we iterate over all the entities within the scope of entity type to
arrive at the sorted set of entities. So, this IMO should definitely be fixed
by providing some sort of index table.
In the 2nd point in my [comment above |
https://issues.apache.org/jira/browse/YARN-5585?focusedCommentId=15494251&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-15494251]
we can query entity ID specific entities directly from entity table.
One more suggestion was to open up an interface which can be used to provide
encoding and decoding for specific Entity IDs' (based on entity type) as part
of row key.
This would not require any extra write or read. However, Li and Rohith seemed
to be a little reluctant with that solution as Tez or Spark will have to add
code for it, albeit only a little bit.
However, as [~vrushalic] suggested we can also create an auxiliary table, and
specify the key in timeline entity. Issue with this is we are sort of exposing
internal implementation.
This however can be useful if we want to sort by something else as well as
pointed out, not merely created time. Problem though can be double write. How
about having this auxiliary table as an index table ? And have one write just
to make an entry into this table.
On read side though we can refer to this index table depending on the
suggestion made by Vrushali i.e. specify the index table and start row key and
then use MultiRowRangeFilter to get records from entity table.
Thoughts ?
However, I do feel we inherently need to support created time based sorting
scenario (i.e. have created time based index table as a mandatory table without
user needing to specify it in REST) as we promise in the interface that
entities will be sorted in that fashion.
Probably we can discuss further on this in call today
> [Atsv2] Add a new filter fromId in REST endpoints
> -------------------------------------------------
>
> Key: YARN-5585
> URL: https://issues.apache.org/jira/browse/YARN-5585
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: timelinereader
> Reporter: Rohith Sharma K S
> Assignee: Rohith Sharma K S
> Priority: Critical
> Attachments: YARN-5585.v0.patch
>
>
> TimelineReader REST API's provides lot of filters to retrieve the
> applications. Along with those, it would be good to add new filter i.e fromId
> so that entities can be retrieved after the fromId.
> Current Behavior : Default limit is set to 100. If there are 1000 entities
> then REST call gives first/last 100 entities. How to retrieve next set of 100
> entities i.e 101 to 200 OR 900 to 801?
> Example : If applications are stored database, app-1 app-2 ... app-10.
> *getApps?limit=5* gives app-1 to app-5. But to retrieve next 5 apps, there is
> no way to achieve this.
> So proposal is to have fromId in the filter like
> *getApps?limit=5&&fromId=app-5* which gives list of apps from app-6 to
> app-10.
> Since ATS is targeting large number of entities storage, it is very common
> use case to get next set of entities using fromId rather than querying all
> the entites. This is very useful for pagination in web UI.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
