I have strange problem with NFS in 16.04.
I'm trying to upgrade my servers (thru reinstalls) from 14.04 to 16.04. The servers access some NFS shares hosted on an OmniOS server with ZFS storage. Everything is virtualized on the same ESXi host and connected to the same vSwitch, giving them 10Gb network internally between them (the limiting factor here should be the drives).
On 14.04 the NFS read speed is okay (> 100 MB/s), but on 16.06 NFS reads grinds to halt (500 kB - 3 MB/sek) within seconds. Wait-for-IO (top) increases to 99.9 %.
The shares are the same and the servers are "identically" set up, regarding disks, storage and network. NFS shares mounted with default options. Write speeds seem okay in both cases.
Sample rsync run from 16.04, read first, then write (had to ^C it after a while):
lch@ub-test:~$ rsync --progress /mnt/multimedia/ubuntu-16.04.1-desktop-amd64.iso . ubuntu-16.04.1-desktop-amd64.iso 374,374,400 24% 456.86kB/s 0:41:32 ^C rsync error: received SIGINT, SIGTERM, or SIGHUP (code 20) at rsync.c(632) [sender=3.1.1] rsync: [sender] write error: Broken pipe (32) lch@ub-test:~$ rsync --progress ubuntu-16.04.1-desktop-amd64.iso /mnt/multimedia/ubuntu-16.04.1-desktop-amd64.iso.ub-test ubuntu-16.04.1-desktop-amd64.iso 1,513,308,160 100% 178.83MB/s 0:00:08 (xfr#1, to-chk=0/1) lch@ub-test:~$
Sample rsync run from 14.04, read first, then write:
lch@utility:~$ rsync --progress /mnt/multimedia/ubuntu-16.04.1-desktop-amd64.iso . ubuntu-16.04.1-desktop-amd64.iso 1,513,308,160 100% 127.14MB/s 0:00:11 (xfr#1, to-chk=0/1) lch@utility:~$ rsync --progress ubuntu-16.04.1-desktop-amd64.iso /mnt/multimedia/ubuntu-16.04.1-desktop-amd64.iso.utility ubuntu-16.04.1-desktop-amd64.iso 1,513,308,160 100% 244.77MB/s 0:00:05 (xfr#1, to-chk=0/1)