qpid-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gordon Sim <g...@redhat.com>
Subject Re: Journal Full Error
Date Mon, 27 Jul 2009 07:51:00 GMT
Carl Trieloff wrote:
> Kim van der Riet wrote:
>> On Fri, 2009-07-24 at 15:13 -0400, Kim van der Riet wrote:
>>>> 2) I thought the flow-to-disk policy would result in excess messages
>>> being accepted into a slower interim storage, not being rejected or
>>> causing the broker to crash.
>>> This is incorrect. However, staged messages (ie messages that exceed a
>>> broker per-message size threshold) are stored outside the journal in
>>> BDB.
>> I take this back; Gordon is correct. Transient messages are in fact
>> stored in BDB.
> Sorry to banter on the user list.
> - Message where the size is larger than  --default-queue-limit BYTES 
> (104857600) go to non journal storage

I believe you mean to refer to the --staging-threshold parameter here; 
this is the threshold at which a large message is written to disk as it 
arrives (messages smaller than this limit are written once the full 
message has been received).

The default-queue-limit sets the max in-memory size for the queue (i.e. 
the cumulative size of all messages); for a durable queue this is the 
point at which further messages will be 'flowed-to-disk'.

> - All other messages, transient or durable if flow_to_disk go to the 
> Journal. (just checked the code on this point)

I believe this is also incorrect. Transient messages that are deemed to 
have exceeded the queue limit of a flow-to-disk queue are only released 
from memory after they are written to disk via the 
'MessageStore::stage()' method which in the journal based store is 
written to a BDB database not the queue's journal.

Apache Qpid - AMQP Messaging Implementation
Project:      http://qpid.apache.org
Use/Interact: mailto:users-subscribe@qpid.apache.org

View raw message