You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@trafodion.apache.org by "Atanu Mishra (JIRA)" <ji...@apache.org> on 2016/02/11 02:26:18 UTC

[jira] [Created] (TRAFODION-1823) ESP idle timeout does not kick in, leading to too many ESPs on the system

Atanu Mishra created TRAFODION-1823:
---------------------------------------

             Summary: ESP idle timeout does not kick in, leading to too many ESPs on the system
                 Key: TRAFODION-1823
                 URL: https://issues.apache.org/jira/browse/TRAFODION-1823
             Project: Apache Trafodion
          Issue Type: Bug
          Components: sql-exe
    Affects Versions: 1.3-incubating
            Reporter: Atanu Mishra


There are two mechanisms to time out idle ESPs:
•	The master executor kills idle ESPs after a timeout, based on session default ESP_IDLE_TIMEOUT. However, this only happens when we allocate another statement that needs ESPs, because the needed logic is triggered in the code to allocate ESPs. Example, to set this timeout to 30 minutes (1800 seconds):

set session default esp_idle_timeout '1800';
•	ESPs also have a built-in timeout a period of inactivity. The same session default ESP_IDLE_TIMEOUT is used as well. This timeout should trigger automatically, no action from the master is required.
Bug: Right now, this second timeout isn't working, because the idle ESP sees two connections from the master, and that is not considered an idle ESP.





--
This message was sent by Atlassian JIRA
(v6.3.4#6332)