[
https://issues.apache.org/jira/browse/ARIA-265?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ran Ziv updated ARIA-265:
-------------------------
Description:
When using multi-threaded plugins, sometimes a {{Database is locked}} error
gets raised.
The reason for this is that each thread automatically receives a separate
session from {{SQLAlchemy}} for communication with the database. This session
must be closed properly before the thread terminates, or else other attempts of
interacting with the database could hang (and lead to the error mentioned
above).
One way to avoid this is for plugin authors to ensure they shut the session
down properly for each thread the plugin creates.
Another possible solution would be to save all sessions created on the
operation's {{ctx}} object, and clean them up automatically afterwards.
was:
When using multi-threaded plugins, sometimes a {{Database is locked}} error
gets raised.
The reason for this is that each thread automatically receives a separate
session from SQLAlchemy for communication with the database. This session must
be closed properly before the thread terminates, or else other attempts of
interacting with the database could hang (and lead to the error mentioned
above).
One way to avoid this is for plugin authors to ensure they shut the session
down properly for each thread the plugin creates.
Another possible solution would be to save all sessions created on the
operation's {{ctx}} object, and clean them up automatically afterwards.
> Database locked errors when using multi-threaded plugins
> ---------------------------------------------------------
>
> Key: ARIA-265
> URL: https://issues.apache.org/jira/browse/ARIA-265
> Project: AriaTosca
> Issue Type: Bug
> Reporter: Ran Ziv
> Priority: Minor
>
> When using multi-threaded plugins, sometimes a {{Database is locked}} error
> gets raised.
> The reason for this is that each thread automatically receives a separate
> session from {{SQLAlchemy}} for communication with the database. This session
> must be closed properly before the thread terminates, or else other attempts
> of interacting with the database could hang (and lead to the error mentioned
> above).
> One way to avoid this is for plugin authors to ensure they shut the session
> down properly for each thread the plugin creates.
> Another possible solution would be to save all sessions created on the
> operation's {{ctx}} object, and clean them up automatically afterwards.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
