Kbase P117832: Fathom Management: 100% CPU metrics on Linux remote containers
| Autor |
Progress Software Corporation - Progress |
| Acesso |
Público |
| Publicação |
12/31/2008 |
|
Status: Verified
SYMPTOM(s):
Fathom Mangement showing 100% CPU usage on remote containers
100% CPU metrics
CPU Stats peak at 100% and remain there
Remote machine OS utilites show CPU at 3 - 5 %
Multi-processor CPU at 100%
Only remote OS metrics on Linux servers are wrong
Fathom AdminServer has been stopped and restarted
Latest Service Packs applied 3.0A04
Latest Service Packs applied 10.0B05
Latest Service Packs applied 9.1E04
FACT(s) (Environment):
Progress 9.1E
OpenEdge 10.0x
Fathom Management 3.0A
Fathom Management 3.1A
Linux Intel
CAUSE:
Bug# 20060801-004
CAUSE:
The problem is related to an overflow when values accumulate over 4GB, which can be verified on the remote system using `cat /proc/stat`. On Linux Fathom CPU disk monitoring is not correctly handling large values (exceeding 4GB) in the /proc/stat file, causing Fathom CPU monitoring on Linux to indicate 100% CPU usage incorrectly.
FIX:
Fathom 3.1A introduces mutli-processor monitoring. This overflow has been fixed in Fathom Management 3.1A02 and 3.1B where interim cpu values are now stored in a 64 bit integer.