ISILON SMB - file in Use error

I have combined a few of the previously mentioned into one that works well for me.

isi_for_array isi smb openfiles list --verbose | grep -A3 "%filename%"

I am constantly unlocking locked files due to an error we are unable to narrow down. I prefer the following:

isi_for_array  "isi smb openfiles list --verbose" |grep -B3 -A3 (Username)

OneFS: MS Office 2010/2013 Excel "file is locked by another user".

If Users are behind a firewall, check the keepalive / keepidle times of the tcp settings on client / firewall and cluster. Sometimes a firewall closes the tcp-session without correctly informing client and server.

Problem with user session, timeouts, and open files.

Yes, keepalive is very important, once I had a case where firewall was configured with 60 sec instead of 60 minutes.

We are struggling to find a solution for this one. We got no firewalls in place between the isilon and the clients. There is no firewall installed on the Windows 7 clients. The error does not happend on every client whicht is a miracle because we use GPOs, Central Software Distribution and Images for Windows Installation, but still the clients behave different.

Allow Trusted Locations and Procteded Views is Setup like you recommended in your Blog.

The error appears when a user locks his workstation with an opened but unsaved document and comes back in after 30 or more Minutes and hits the safe button.

Ingo

We are unable to perform an tcpdump on isilon, this cluster is way to slow for this. I cannot see anything withing this trace pointing to an error. The SR is open for a few months but still without a solution.

I even applied your solution last tuesday, but again without any improvements.

the cluster is too slow to run tcpdump ?  How's that ?

I got four X200 Nodes each with 10 1TB SATA and a single SSD having 7500 SMB connections. There is diskless A100 as well (but without any use, EMC overlooked that our rusty VTL is not listed in their HCL). Whenever I start a capture the node stops responding to any new connection. With an clusterwide tcpdump the whole cluster stales.

From my point of view the sizing is way to small. I made a presentation showing up the sizing is to small. They expected to push out 12.000 Protocol Request with the nodes w. 11 1TB SATA drives each. SyncIQ to the secondary site is running as well. They agreed and we added a ssd of the delivered nodes and another node to each site.

Ingo

Ingo

tcpdump troubles even if filtering traffic from/to a single test client?

Have you considered using the idle A100 and a dedicated address pool

for testing/tcpdump with selected clients? (Some might say using Backup-A100

for NAS traffic is not recommend, but should be fine for isolated test cases.)

-- Peter

If you see "file in use" error and network trace shows STATUS_SHARING_VIOLATION  and your cluster version is 7.0.2.4 -7.0.2.9 : with high chance you have to upgrade to OneFS 7.0.2.10.