Devesh Kumar Singh created HDDS-13515:
-----------------------------------------
Summary: Recon - DB checkpoint based processing by Recon during
full OM DB snapshot based bootstrap
Key: HDDS-13515
URL: https://issues.apache.org/jira/browse/HDDS-13515
Project: Apache Ozone
Issue Type: Improvement
Components: Ozone Recon
Reporter: Devesh Kumar Singh
Assignee: Devesh Kumar Singh
During full snapshot fetch of OM DB by Recon , ReconOmTasks cleans or truncate
the existing table data. Same way other task classes handles insert update
accordingly. Now the problem with this approach when Recon falls back on full
snapshot and then we do cleanup of existing processed data from their
respective task tables, Recon APIs used by Recon UI will not return any data
and Recon UI will show all blank , this makes bad experience for user, so We
need a new solution because when Recon falls back on full snapshot, then Recon
will do reprocess , but we need a mechanism where existing data is not cleaned
and Recon reprocess should work on a checkpoint of DB as a input when full OM
DB is fetched and output a new staging DB to store those ReconOMTask tables
like nsSummaryTable till reprocess of all ReconOMTask classes gets over and
then this staging DB should be moved or overwritten to a final DB from where
Recon APIs will read data finally.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
