tomee-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jlemoine <>
Subject Re: HTTP Errors on Load-Balanced TomEE Servers
Date Mon, 03 Oct 2016 15:11:51 GMT
Romain Manni-Bucau wrote
> 2016-10-03 15:43 GMT+02:00 Jlemoine &lt;

> jlemoine@

> &gt;:
>> 1.  I'm not sure what a "very small timeout" would be.  I currently have
>> the
>> ajp connector and LB timeouts at 20 seconds.
> 1ms?
>> 2.  Yes, ajp does seem to keep threads open once created, which is where
>> I
>> think the problem was.  No new sockets available to open.  However, so
>> far
>> this has not happened on the TomEE side after I changed the LB method to
>> Busyness and made the other changes to the ajp connectors that I
>> mentioned
>> in the first post.
> this is weird, ajp perf boost comes normally from the binary protocol
> usage
> and connections staying open. Depends the setup for sure but if not used
> wonder if you have any interest to use ajp.
>> --
>> View this message in context: http://tomee-openejb.979440.
>> tp4680255p4680272.html
>> Sent from the TomEE Users mailing list archive at

Yes Romain, from what I've researched what you're saying holds true.  But I
had the ajp attributes at default and this bizarre behavior from the LB
Request method was totally overloading an instance.

This isn't really load balancing for us and any end-users attached to that
"a" instance were locked up and unable to process data back to the database.
I'm pretty sure I have this ironed out now but I won't know unless the
Business wants to try out my config changes in Production.  Also, I'm still
researching if there are other tuning options that I may need to try.

View this message in context:
Sent from the TomEE Users mailing list archive at

View raw message