ENS 10.6.1 December 2019 update - memory leak

[ezim]

We are having issues on our file servers with the On-Demand Full scans since upgrading to the ENS 10.6.1 December 2019 update.

It looks like the memory allocated during the scan rises, but does not get returned once the scan finishes.
This leads to the servers running out of memory during or shortly after the full scans and crash.

Has anybody else experienced these issues?

It happens on servers running various versions of Windows OS.
They all have large amounts of data to scan.
They were all updated to ENS 10.6.1 December 2019 patch.

[Former Member · Accepted solution]

Hi @ezim 

There have been a couple cases with this symptom opened through the service portal, but we have yet to receive the proper dataset that we need in order to fully investigate the problem. That being said, preliminary investigation based on the incomplete datasets has indicated that this might be related to an issue fixed in the February releases of ENS.

Could you please try installing the 10.6.1 February (2020) update on one system experiencing the issue, rebooting the system, then performing an ODS and seeing if you can reproduce the issue?

If the issue re-occurs with the February update, please create an SR and PM me the SR number so I can work with the SR owner to get you data collection instructions relevant to your environment.

Thanks,

[ezim]

Hello @Former Member ,

Thank you for confirming that there appears to be an issue.

I won’t be able to install the February update until later in the week.

You mention that you don’t have a proper dataset. What information do you still require?

Thank you.

[Former Member]

Hi @ezim 

Generally speaking, we need the below items:

1. System configured for a full memory dump, and to allow for a keyboard crash of the system
2. gflags enabled on system (all flags, including kernel flags)
3. WPR trace captured for 30-60 seconds while the handle leak is occurring during an ODS
4. Full memory dump triggered by a keyboard crash once the WPR trace has been saved

But the above data need only be captured if the issue is still observed with the February 2020 update.

Thank you,

[Former Member]

I am working with Microsoft to identify this issue and we were trying to get a WPR trace but if we let the ODS run for anymore than a few minutes the WPR won't run.  I don't know how you're going to get that information.

[Former Member]

Hi @Former Member 

We've seen this issue resolved by installing the ENS 10.6.1 or 10.7.0 February update. Rather than collecting this data, which is difficult to perform due to the resource exhaustion on the system, would you please test this version of the product to confirm it resolves the issue for you as well?

Thanks,