I'm trying to use OpenSolaris as a zfs-over-nfs file server. I have one client, a Mac, connected to my Solaris box via a linksys router. The host and client are assigned static IP addresses. (My system/OS info is listed at the end of this post.)
I can mount the NFS directory on my Mac and begin copying files over to the server. I am copying massive amounts of files, so I take off to get coffee, go to work, whatever. Each time I come back and check the status of the copy, I find the connection was disrupted, typically after 4-5 GB had been copied. The Mac system log shows the following: May 29 14:45:08 AJWlaptop kernel[0]: nfs server 192.168.1.100:/mus: not responding May 29 14:45:08 AJWlaptop kernel[0]: nfs server 192.168.1.100:/mus: not responding May 29 14:45:08 AJWlaptop kernel[0]: nfs server 192.168.1.100:/mus: not responding May 29 14:45:08 AJWlaptop KernelEventAgent[133]: tid 00000000 received VQ_NOTRESP event (1) May 29 14:45:08 AJWlaptop KernelEventAgent[133]: tid 00000000 type 'nfs', mounted on '/Volumes/192.168.1.100', from '192.168.1.100:/mus', not responding May 29 14:45:08 AJWlaptop KernelEventAgent[133]: tid 00000000 found 1 filesystem(s) with problem(s) May 29 14:45:39 AJWlaptop kernel[0]: nfs server 192.168.1.100:/mus: not responding May 29 14:45:39 AJWlaptop kernel[0]: nfs server 192.168.1.100:/mus: not responding May 29 14:45:39 AJWlaptop kernel[0]: nfs server 192.168.1.100:/mus: not responding ... May 29 14:47:12 AJWlaptop kernel[0]: nfs server 192.168.1.100:/mus: not responding May 29 14:47:13 AJWlaptop KernelEventAgent[133]: tid 00000000 unmounting 1 filesystems Accompanying the dropped connection is a Solaris system freeze -- the keyboard and mouse are unresponsive. At first I thought this was a power management issue; some process was putting the machine into idle and disrupting the server. So I tried all of the following: * svcadm disable system/power * adding "autopm disable" to /etc/power.conf and rebooting * I even tried disabling acpi by adding "setprop acpi-user-options '0x2'" in /boot/solaris/bootenv.rc and rebooting I also made sure the versions of NFS were compatible; I set both machines to use nfsv3 (the highest version available on my Mac). No matter what I tried, the system still froze and dropped the connection after a time. Then, while I was trying another copy to the server (~20GB of files) and typing furiously at my Solaris box to try to diagnose the problem, I noticed that the copy had gone beyond the previous 5 GB limit; I kept typing periodically for the duration of the copy, and it finished without a freeze (after 3 hours). It seems that there is some process putting the computer into a lower power state when the keyboard is idle for x amount of time. But I can't figure out what process is causing this. I killed powerd with the commands issued above. Could it be the motherboard? There is not a way to disable ACPI in the BIOS. My system info: Solaris: SunOS Release 5.10 Version Generic_127128-11 64-bit / AMD64 processor / Gigabyte GA-MA770-DS3-E mohterboard / 1 GB RAM Mac: Mac OS X v. 10.4.11 PowerPC Any help is greatly appreciated; I've been trying to figure this out for a couple of days and have invested a lot in the idea of opensolaris/zfs. Best, Adam This message posted from opensolaris.org
