Hello,
I have been struggling with this for hours/days/weeks. Tried many different approaches to searching google and the vmware forums. But I'm not finding an answer. I hope someone can provide me with some advice on where to find an answer.
I've got 2 ESXi 5.0.0 hosts configured in vSphere Essentials (vCenter). Each host has its own dedicated network storage device (ReadyNAS 2100) configured with an NFS export. These NFS exports are used to store the VM disk files on them. Both host and ReadyNAS devices are configured identical as far as I can tell.
For some reason, one of the ESXi hosts is generating tons of read requests per second against the ReadyNAS at various times throughout the day/night. It does not seem that any of the VM's on the host are generating these read requests. They are generated by the host. This flood of read requests usually lasts multiple hours and saturates the storage network with 50+mb of traffic consistently.
In doing a network dump on the host, I can see all the NFS read request data coming across the network, and at the time, most of it seemed to be a bunch of mail spooler data.
Is there some sort of sync process that could be running on the ESXi host? Is there a script/program I can run on the host to determine what it is actually doing?
The only obvious difference between the hosts, is that the host having troubles has a VM with old mbox files on it that are multiple GB's in size, whereas all other VMs use the newer maildir format where each email is its own separate file. Several times throughout the day, all VM's on the problem host have latency issues. Once I power down the VM with mbox formatted email, everything immediately speeds up again. Though this doesn't seem to have any effect on the 50+mb of traffic that the ESXi host is receiving from its read requests against the NFS datastore.
Any advice would be greatly appreciated.
Thank You