Bug 622865 - QMF agent should throttle back connection and object updates when queue full detected.
Summary: QMF agent should throttle back connection and object updates when queue full ...
Status: CLOSED WONTFIX
Alias: None
Product: Red Hat Enterprise MRG
Classification: Red Hat
Component: qpid-qmf
Version: beta
Hardware: All
OS: Linux
low
high
Target Milestone: Next Version
: ---
Assignee: Ken Giusti
QA Contact: MRG Quality Engineering
URL:
Whiteboard:
Keywords:
Depends On: 623805
Blocks: 615441
TreeView+ depends on / blocked
 
Reported: 2010-08-10 15:30 UTC by Ken Giusti
Modified: 2011-08-12 16:04 UTC (History)
2 users (show)

(edit)
Clone Of:
(edit)
Last Closed: 2011-03-10 14:24:05 UTC


Attachments (Terms of Use)
Backoff reconnect attempt if connection dropped due to queue limit violation. (3.91 KB, patch)
2010-08-12 17:38 UTC, Ken Giusti
no flags Details | Diff

Description Ken Giusti 2010-08-10 15:30:07 UTC
Description of problem:

The QMF agent effectively ignores queue resource exhaustion events.  Currently, the agent will immediately reconnect and continue publishing data when a queue exhaustion exception occurs.  This behavior exacerbates the queue saturation problem by creating additional unsolicited management messages (binding, etc).

Version-Release number of selected component (if applicable):
Beta

How reproducible:
100%


Steps to Reproduce:
1. See https://bugzilla.redhat.com/show_bug.cgi?id=615441
2.
3.
  
Actual results:
The QMF agent immediately re-connects and continues publishing object status messages.

Expected results:
The agent should handle the exception more gracefully by backing off the reconnection attempt, and further by delaying the next object update cycle/throttle back the number of objects published.

Additional info:

Comment 1 Ken Giusti 2010-08-12 17:38:52 UTC
Created attachment 438490 [details]
Backoff reconnect attempt if connection dropped due to queue limit violation.

I've tried the above patch in testing in my high traffic/slow consumer setup - there was no noticible improvement to performance when simply backing off the reconnect attempt.   Including patch for reference - not submitting it as a fix.


Note You need to log in before you can comment on or make changes to this bug.