Unsolved
This post is more than 5 years old
4 Posts
0
1213
PerformancePoll on EV_ExpandableSubsystem
Hello *,
post a disk replacement done on customers system (VNX5600, ten minutes prior to that event), and timely fitting to an outage starting at that time, I've seen the following lines in the "getlog" output:
| 12/31/2015 00:54:05 N/A (4600)'Log In' called by ' Navi User admin' (10.19.14.28) on 'Security Service' (Result: Success). User has logged in. | |
| 12/31/2015 00:54:47 N/A (4600)'Poll' called by ' Navi User admin' (10.19.14.28) on 'EV_ExpandableSubsystem' (Result: Success). | |
| 12/31/2015 00:54:48 N/A (4600)'PerformancePoll' called by ' Navi User admin' (10.19.14.28) on 'EV_ExpandableSubsystem' (Result: Success). | |
| 12/31/2015 00:54:49 N/A (4600)'PerformancePoll' called by ' Navi User admin' (10.19.14.28) on 'EV_ExpandableSubsystem' (Result: Success). | |
| 12/31/2015 00:55:21 N/A (4600)'PerformancePoll' called by ' Navi User admin' (10.19.14.28) on 'EV_ExpandableSubsystem' (Result: Success). |
All my attempts to reproduce that on our testbed failed, even though I've seen such lines also on CX4-120.
Can someone explain the meaning and whether it remarks something critical?
Thanks in advance
Frank
kelleg
4.5K Posts
0
January 6th, 2016 10:00
Those look like a user (admin) logging in to the array from the IP address (10.19.14.28). That could be a host on the network or the Control Station if you have a Unified VNX (the FILE side). It looks like a polling event asking from performance data from the array. See if you can find out what that IP address is.
glen
ZaphodB
195 Posts
0
January 6th, 2016 11:00
Is Monitoring and Reporting running in the environment? I would expect to see something like this from the naviseccli activity that the M&R host does to collect stats.
You would not see it in a system that is not registered in M&R, so that might explain why your testbed doesn't see it.
elmigratione
4 Posts
0
January 7th, 2016 00:00
Hello *,
thanks for the reply.
@glen: as mentioned, these messages were seen post a hardware replacement. The IP address belongs to the machine the EMC technician was using for the maintenance activity.
I also think that a kind of polling request is trying to retrieve performance data, but how got this triggered? And does this potentially impact overall box behavior in the way, that an Oracle instance (running in RAC on top of ASM) is undergoing an unwanted warm restart?
@Zaphod: M&R are not running.
Thanks®ards
Frank
kelleg
4.5K Posts
0
January 7th, 2016 09:00
The CE was probably running spcollects from the laptop - spcollects also collect performance data when run. This would normally not impact the performance on the array unless the array is already heavily utilized, then running spcollects could potentially cause some latency that an application might detect, but that should not cause Oracle to restart. Spcollects are run all the time on arrays without causing this issue unless the array is already overloaded.
You didn't mention what OS is running on the hosts, but some types of disk errors can cause extra latency that can be seen by Oracle if you're running Redhat and PowerPath. We've seen issues with the version of Redhat and PowerPath that can cause a disk failure to affect Oracle (timeouts that drop connections and Oracle stops as all paths are marked incorrectly as dead). If you are running Redhat/PowerPath, make sure that the versions are correct.
glen