James 2.1.2 - message stuck in transport state; spool grows; no error messages

[Robert J Taylor <ro...@rjamestaylor.com>]

Hi.

While anticipating 2.2.0 were are still using 2.1.2 with MySQL in
production. In the past month we have experienced a "silent sticking" of
messages in the spool.

I cannot see a message in the logs or going to STDERR (I log STDERR,
too, just in case) that relates to the stuck and growing spool queue. In
fact the way I know the condition exists is that email is not being
delivered (and I now monitor the spool row count).

Example, today.

mysql> select message_state, count(*) from spool group by message_state;

message_state, count(*)
-------------------------
root, 3
transport, 101


Previous days we copy the spool, delete trash messages (somehow related
it seems, though no message says so), shutdown James via remotemanager,
delete spool, insert parred queue and restart. Sometimes this process
takes hours. (There has to be a better way!).

Today we set up a second instance of James without POP, SMTP, etc,  and
allowed it to hit the spool. On the first start up one message escaped
the spool. Then, nothing. We shutdown the second instance and restarted
it. A few more messages escaped. Now the spool queue looks thusly:

mysql> select message_state, count(*) from spool group by message_state;

message_state, count(*)
-------------------------
error, 2
root , 8
transport, 90


Shutdown instance 2, restart it, and this:

mysql> select message_state, count(*) from spool group by message_state;

message_state, count(*)
--------------------------
root, 10
transport, 89


Our configuration, modified judiciously, is included below. All help
appreciated.


Thanks,

Robert Taylor
robert@rjamestaylor.com

Re: James 2.1.2 - message stuck in transport state; spool grows; no error messages

[Robert J Taylor <ro...@rjamestaylor.com>]

Noel,

You nailed it. There I was *increasing* threads to break out of what 
appeared to be single-threaded behavior. I hadn't considered 
deadlocking.  Immediate problem solved.

Thanks.

Robert Taylor

Noel J. Bergman wrote:

>>While anticipating 2.2.0 were are still using 2.1.2 with MySQL in
>>production. In the past month we have experienced a "silent
>>sticking" of messages in the spool.
>>    
>>
>
>There was a deadlock found in the database connection code in 2.1.3.  It was
>probably in 2.1.2 as well.  Also, there was a change between 2.1.2 and 2.1.3
>that might have effected what you are seeing.
>
>If you cannot update to 2.2.0 RC2 (seems that you'd have little to lose,
>given the problems you're encountering now), try setting the number of spool
>threads to 1 and see if that has an effect.
>
>	--- Noel
>
>
>  
>

RE: James 2.1.2 - message stuck in transport state; spool grows; no error messages

["Noel J. Bergman" <no...@devtech.com>]

> While anticipating 2.2.0 were are still using 2.1.2 with MySQL in
> production. In the past month we have experienced a "silent
> sticking" of messages in the spool.

There was a deadlock found in the database connection code in 2.1.3.  It was
probably in 2.1.2 as well.  Also, there was a change between 2.1.2 and 2.1.3
that might have effected what you are seeing.

If you cannot update to 2.2.0 RC2 (seems that you'd have little to lose,
given the problems you're encountering now), try setting the number of spool
threads to 1 and see if that has an effect.

	--- Noel


---------------------------------------------------------------------
To unsubscribe, e-mail: server-user-unsubscribe@james.apache.org
For additional commands, e-mail: server-user-help@james.apache.org