[ 
https://issues.apache.org/jira/browse/ATLAS-2092?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16148957#comment-16148957
 ] 

Nigel Jones commented on ATLAS-2092:
------------------------------------

thanks for the very thorough write-up and walk through.

In terms of resolution a little thinking 'out-loud'

* At the point we switch to janus and drop titan 0.54 we ideally don't want any 
workarounds in that version of the code, but we may need to make sure we 
address any odd things in the graph during a migration step?
* You mention that fixing titan 0.54 is unrealistic. I understand there is no 
longer any support/activity, it's a dead end. However if it is a small, precise 
change that is required, I still wonder if could build a patch that can be 
applied to 0.54, and ship it with atlas. We could also ensure the libraries we 
supply have it included (needs a build), and perhaps detect on startup, and log 
an error if patch isn't in place. Though this is a bit tedious, I still wonder 
if it's safer or more controlled than adding a fair bit of workaround code. may 
be worth thinking about?
* However in the case of any fix (as in previous point) am I right in thinking 
once this has occurred it may continue to cause functional issues since the key 
will always be out of the correct range? In which case we'd still either need 
to be able to detect, and/or self-heal/fix on demand, or have a fix-script 
available for old atlas users? this may make my previous point somewhat mute.

> Failures following concurrent updates
> -------------------------------------
>
>                 Key: ATLAS-2092
>                 URL: https://issues.apache.org/jira/browse/ATLAS-2092
>             Project: Atlas
>          Issue Type: Bug
>          Components:  atlas-core
>            Reporter: Graham Wallis
>         Attachments: Investigations and findings relating to concurrent 
> updates in Atlas.pdf
>
>
> There is a race condition that causes duplication of schema vertices as a 
> result of concurrent graph updates. This in turn leads to failure of queries 
> that specify a type such as an edge label used in an attribute that 
> references another entity. This problem is known to affect Atlas entity refs 
> – which create graph edges that use edge label schema vertices. It is likely 
> that it also affects other types in Atlas.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Reply via email to