My Oracle Support Banner

How Will CPU Usage By Flag Updates Behave With Casandra Store? (Doc ID 2428519.1)

Last updated on AUGUST 02, 2018

Applies to:

Oracle Communications Messaging Server - Version 8.0.2 and later
Information in this document applies to any platform.

Goal

We had an issue where a user had a very large number of concurrent connections and it caused high CPU usage.  The cause of the high CPU usage was due to the inefficiency of the flag update scan on very large folders, exacerbated by the user having many IMAP sessions with the large folder selected, **AND** a steady stream of new mail arriving in that large folder.

The load average went from normally around 10-20 to over 300. Various monitoring tasks were starting to generate alerts: because the load was so high, but also because the load being so high was causing the checks to timeout.  We were able to identify a top user via Splunk and confirmed the user had hundreds of IMAP connections open and selected a folder which had 3 million messages and that folder was steadily receiving new messages.

imsconnutil -k -u ; immediately returned the load average to normal. But the user's various client devices also immediately started reconnecting and the problem returned quickly.

We finally set the user's mailUserStatus=defer and mboxutil -r to rename that folder to an _old name, which took quite some time.

Q1. Is there anything else that can be fixed in the code to reduce the impact of this sort of thing?
Q2. How does the Casandra store handle this sort of situation?
 

Solution

To view full details, sign in with your My Oracle Support account.

Don't have a My Oracle Support account? Click to get started!


My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.