The uxioserv uses almost 100% of the cpu when the system time changes

Document ID : KB000087105
Last Modified Date : 14/04/2018
Show Technical Document Details
Issue:
Error Message :
No error appeared at all on the logs, but when launching uxiotrclck during the issue, we observed that this kind of message appeared all the time many times on the generated IO_DATE.dump.log

###########
|TRACE|X|IO |pid=3945.3004165008| o_io_thread_module | engine tid:-1290802288 tempo:0
########### 

Patch level detected:Dollar Universe 6.4.22
Product Version: Dollar.Universe 6.4.22

Description :We realized that the process uxioserv used a lot of cpu at certain time and that could only be fixed after stopping / starting the node.
The cpu usage was almost 100% even when we stopped all the engines except the IO.
Environment:
OS: Linux
Cause:
Cause type:
Configuration
Root Cause: The issue was due to the fact that the NTP and the VMTools had a different time reference server, with a difference of around 61 seconds which caused that the system time was constantly switching back and forth.
Resolution:
The issue was fixed after stopping the vmtools which had an incorrect reference time.
The real solution is changing the ESX time so it is synchronized with the NTP reference server.

Fix Status: Released

Fix Version(s):
Component: Application.Server
Version: Dollar.Universe 6.4.22
Additional Information:
Workaround :
N/A