Pillar Axiom: High NFS activity can cause slammer failover failback

Asset ID:	1-72-1437287.1
Update Date:	2012-08-29
Keywords:

Solution Type Problem Resolution Sure

Solution 1437287.1 : Pillar Axiom: High NFS activity can cause slammer failover failback

Applies to:

Pillar Axiom 300 Storage System - Version Not Applicable and later
Pillar Axiom 500 Storage System - Version Not Applicable and later
Pillar Axiom 600 Storage System - Version Not Applicable and later
Information in this document applies to any platform.

Symptoms

An Axiom can undergo a failover, then failback successfully. A sample case study is below:

* Activity leading up to this event can be due to a barrage of tests being performed to harden their security (which involves Penetration testing (credentialed)) where they're logged into different clients with NFS access, where there's tons of SSH logins and heavy-access activity following immediately thereafter. They experienced where some hosts developed file inaccess as a result.

Cause

Under certain circumstances if the NFS client is extremely busy and not acknowledging the response from the AXIOM, it is possible that the continuously growing resources tied to the client can result in resource starvation for other clients. This could result in increased latency in the AXIOM services and the AXIOM could warmstart to recover.

The Axiom runs out of zero copy buffers which is currently set at 40k, and can be due to memory leak or heavy read requests

Solution

An upgrade to a release above 040500 is recommended where we may place additional checks to provide diagnostics back to engineering in case this event happens.

Attachments

This solution has no attachment