[
https://issues.apache.org/jira/browse/ATLAS-503?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Hemanth Yamijala updated ATLAS-503:
-----------------------------------
Attachment: ATLAS-503.patch
Attaching a patch that implements some retries while acquiring locks within a
JVM. The number of retries is configured using
{{atlas.graph.storage.lock.retries}}. As described above, the value for this
property should be equal to the desired level of parallelism. For e.g. if this
is set to 10, it means that 10 concurrent operations can proceed to acquire
locks on the same store, key and column.
Clearly it is difficult to predict this number, nor set it to a very high
value, but until we determine a better solution in terms of fixing the model,
this seems to be the best workaround for now. I have tested this by setting the
value to 10, and creating tags in parallel using 10 concurrent users in JMeter.
Verified that all the queries from the 10 users finished successfully. I will
take [~ssainath]'s help to test this at scale and also to verify we are not
seeing any major regressions with it using current tests.
> Not all Hive tables are not imported into Atlas when interrupted with search
> queries while importing.
> -------------------------------------------------------------------------------------------------------
>
> Key: ATLAS-503
> URL: https://issues.apache.org/jira/browse/ATLAS-503
> Project: Atlas
> Issue Type: Bug
> Reporter: Sharmadha Sainath
> Assignee: Hemanth Yamijala
> Priority: Critical
> Fix For: 0.7-incubating
>
> Attachments: ATLAS-503.patch, hiv2atlaslogs.rtf
>
>
> On running a file containing 100 table creation commands using beeline -f ,
> all hive tables are created. But only 81 of them are imported into Atlas
> (HiveHook enabled) when queries like "hive_table" is searched frequently
> while the import process for the table is going on.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
