[OmniOS-discuss] esxi 5.5 to omnios r151014 nfs server issue

[Richard Elling]

> On Apr 10, 2015, at 8:59 AM, Schweiss, Chip <chip at innovates.com> wrote:
> 
> On Fri, Apr 10, 2015 at 10:23 AM, Richard Elling <richard.elling at richardelling.com <mailto:richard.elling at richardelling.com>> wrote:
> 
> It is my understanding that ESX doesn't use NFS locking, they implement their own.
> This makes sense, because NFSv2/3 locking has been problematic since the beginning,
> from all vendors. Again, this is easy to measure.
> 
> I thought that too, but looking in the advanced config, there are a couple tunables for nfs locks.  NFS.LockRenewMaxFailureNumber, range 1-100, def 3 and NFS.LockUpdateTimeout, range 1-8 default 5.  This is on ESXi 5.5.
> 
> I don't have an non-production datastores connected to ESXi, or else I would test this right away.   

There are several ways to see lock activity on the server side: tcpdump, snoop, dtrace, etc.
There is also kstats for lockd activity on some distros... but offhand I'm not sure exactly what
is in which distro... if we're missing interesting stats we can add them to the open source lock
manager :-) Meanwhile, the other tools can work.
 -- richard

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://omniosce.org/ml-archive/attachments/20150410/43a249b7/attachment.html>