Received-SPF: pass (google.com: domain of linux-nfs+bounces-2135-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.80.249 as permitted sender) client-ip=147.75.80.249;
Precedence: bulk
MIME-Version: 1.0
From: Zhitao Li <zhitao.li@smartx.com>
Date: Fri, 1 Mar 2024 18:47:01 +0800
Message-ID: <CAPKjjnqJZnZb3ja_HgzF7Qzxppnt5E5bsLgL5JzRjw0uDvvVTw@mail.gmail.com>
Subject: Problem: nfstest_cache acdirmin_data/acdirmax_data failures [nfstest] [NFS]
To: mora@netapp.com
Cc: linux-nfs@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"

Hi, mora,

I'm using NFSTest, which helps us a lot.

I met with failures when running nfstest_cache in cases
"acdirmin_data" and "acdirmax_data". The error message is as follows:
 FAIL: Directory listing should have not changed at t=0
 FAIL: Directory listing should have not changed just before acdirmin
 PASS: Directory listing should have changed just after acdirmin

My environment is as follows:
1. Linux kernel for NFS client: v6.7.0
2. NFSTest version: 3.2
3. Test command: nfstest_cache --client xxx --server xxx --export /ns1
--nfsversion=3 -m /mnt/test --datadir=data --runtest=acdirmin_data


Both above cases have the same pattern:
1. Client 1: List some directory.
2. Client 2: add new dentry by mkdir() to the directory
3. Client 1: List the directory.

The cases expect that Client 1 will use directory dentries cache in
Step 1, and will not see the new dentry created by Step 2.  After some
time, client 1 will sync with the NFS server, and see the new dentry.

However, in my environment, client 1 can see the new dentry at once
after client 2 mkdir().  It seems that  the NFS client can check if
the directory has changed and get the latest entries.


Could you give me some help?  BTW, is there any instructions I can
follow to make some contributions?

Looking forward to your early reply.

Best regards,
Zhitao Li.