While debugging some memory issues with out application, we noticed that our RoutingService was reporting an ever-increating value for its periodic "physical memory" metric.
Figured we were configuring something wrong (with history, or lifespan, etc.)
But, if I run up a fairly plain RoutingService via the RTI Launcher (based off of the "DefaultDomainRoute" included in the "RTI_ROUTING_SERVICE.xml" shipped with RTI), I still see this metric increasing.
What's (possibly) even more interesting is that there are no other DDS applications running, other than Admin Console. So there shouldn't be any actual data being routed through the RoutingService.
Screenshots, and XML provided below:
<?xml version="1.0"?>
<dds xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:noNamespaceSchemaLocation="../../resource/schema/rti_routing_service.xsd">
<routing_service name="default">
<annotation>
<documentation>Routes all topics from domain 0 to domain 1</documentation>
</annotation>
<administration>
<domain_id>0</domain_id>
</administration>
<monitoring>
<domain_id>0</domain_id>
<status_publication_period>
<sec>5</sec>
<nanosec>0</nanosec>
</status_publication_period>
<statistics_sampling_period>
<sec>1</sec>
<nanosec>0</nanosec>
</statistics_sampling_period>
</monitoring>
<domain_route name="DefaultDomainRoute">
<participant name="domain1">
<domain_id>0</domain_id>
</participant>
<participant name="domain2">
<domain_id>1</domain_id>
</participant>
<session name="DefaultSession">
<auto_topic_route name="RouteAll">
<publish_with_original_info>true</publish_with_original_info>
<input participant="domain1">
<creation_mode>ON_DOMAIN_AND_ROUTE_MATCH</creation_mode>
<allow_topic_name_filter>*</allow_topic_name_filter>
<allow_registered_type_name_filter>*</allow_registered_type_name_filter>
<deny_topic_name_filter>rti/*</deny_topic_name_filter>
</input>
<output participant="domain2">
<creation_mode>ON_DOMAIN_AND_ROUTE_MATCH</creation_mode>
<allow_topic_name_filter>*</allow_topic_name_filter>
<allow_registered_type_name_filter>*</allow_registered_type_name_filter>
<deny_topic_name_filter>rti/*</deny_topic_name_filter>
</output>
</auto_topic_route>
</session>
</domain_route>
</routing_service>
</dds>
Hi John,
We were able to reproduce the problem. However, we would like to understand how this is affecting your system? Is this issue blocking anything?
--Howard
Hey Howard,
We are troubleshooting some memory issues with a container that is running one or more RoutingServices, along with a couple of other applications.
Trying to discern if and where we might have a mem leak in our apps, but in looking at the various resource graphs, we noticed that upward trend with the RoutingServices.
Wanted to bring it to your attention and to see if it was "just us" or not, or if we are configuring something incorrectly.
So. Our memory issue is a high / blocking issue. But it may or may not be with the RS. Still investigating.
Hi John,
OK, so we have a Jira issue to investigate, but from a scheduling perspective, it's not something that we're expecting to do real soon unless a customer is pushing for it. And the best way to do that is to file a bug with RTI support...when this does become an issue for which you need RTI's attention.
FYI, this forum is usually for users who don't have support and can help each other with simple questions or when RTI folks can help answer "on their own time". For customers with a support contract, it's usually more efficient to just file a query with support.
Howard,
Yup. That was my thinking/understanding as well. Wasn't sure if it warrented a full support ticket or not (since still not sure that we don't have a memory leak on our end).
But, figured it fit the bill of "does anyone else see this as well" kind of question for the forum. :)
Also, if it Was a configuration issue on our end, there was a chance someone would chime in with what I was doing wrong.