[ 
https://issues.apache.org/jira/browse/FALCON-1060?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14358069#comment-14358069
 ] 

Ajay Yadava edited comment on FALCON-1060 at 3/12/15 4:08 AM:
--------------------------------------------------------------

[~svenkat] From what I read, titan doesn't handle the rollbacks. On the 
contrary, it seems from the documentation, titan db lists it as a common gotcha 
^1^.  

Blueprint also lists open transactions as a potential cause which can cause OOM 
exception ^2^ (problem that we are facing) and possible deadlocks as well( 
problem [~kawaa] is facing) which makes us suspect transaction handling in 
graph db even more. Also, I read that a .shutdown() automatically commit all 
transactions and that can possibly explain the recovery by restart which 
[~kawaa] is suggesting.

I might be wrong as I am completely new to Graph Databases and we will try to 
reproduce this with some test cases etc. but it will be very helpful if you 
could provide any links to the docs suggesting otherwise.

References:
1. https://github.com/thinkaurelius/titan/wiki/Transaction-Handling#gotchas 
2. https://github.com/tinkerpop/blueprints/wiki/Graph-Transactions
{quote} Also note, that keeping transactions open for a long time may result in 
OutOfMemoryException if too many mutations have occurred and possible 
dead-locks if locks are held for too long in multi-user environments. {quote}




was (Author: ajayyadava):
[~svenkat] From what I read, titan doesn't handle the rollbacks. On the 
contrary, it seems from the documentation, titan db lists it as a common 
gotcha^1^.  

Blueprint also lists open transactions as a potential cause which can cause OOM 
exception^2^ (problem that we are facing) and possible deadlocks as well( 
problem [~kawaa] is facing) which makes us suspect transaction handling in 
graph db even more. Also, I read that a .shutdown() automatically commit all 
transactions and that can possibly explain the recovery by restart which 
[~kawaa] is suggesting.

I might be wrong as I am completely new to Graph Databases and we will try to 
reproduce this with some test cases etc. but it will be very helpful if you 
could provide any links to the docs suggesting otherwise.

References:
1. https://github.com/thinkaurelius/titan/wiki/Transaction-Handling#gotchas 
2. https://github.com/tinkerpop/blueprints/wiki/Graph-Transactions
{quote} Also note, that keeping transactions open for a long time may result in 
OutOfMemoryException if too many mutations have occurred and possible 
dead-locks if locks are held for too long in multi-user environments. {quote}



> Handle transaction failures in Lineage
> --------------------------------------
>
>                 Key: FALCON-1060
>                 URL: https://issues.apache.org/jira/browse/FALCON-1060
>             Project: Falcon
>          Issue Type: Improvement
>            Reporter: Ajay Yadava
>            Assignee: Ajay Yadava
>
> Transaction failures should be handled on lineage to avoid issues like memory 
> leaks etc.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to