I am building several zfs fileservers using FreeBSD 10, eventually with the goal of getting a pair of primary/replica pairs working from some used Dell 2950's with SAS/6 controllers. I have one pair operational, and was working on the third. The primary question I'm asking is "what's causing and how do I fix the 'bad exports' and related errors".
I start a replica with these commands:
Then every 5 minutes, run the mirror script (inside a locker wrapper "just in case" a replica takes too long; they're running about 20 sec typically, up to a couple minutes occasionally):
These are still a work in progress and the latter hasn't been parameterized yet, and in fact the initialize script hadn't been either, which led to the problem: I initialized the third server as another replica, and in doing so, rotated the repl snapshot so the incremental to the first replica could no longer be done. I grumbled at myself and then re-initialized the first replica, however now when I run the replication, I get these errors on the replica server:
/etc/exports is a 0 length file (it's not used, but the system kept complaining if it wasn't there)
/etc/zfs/exports is, for all the filesystems on vol:
It hasn't changed since I got the system setup in the first place and the replication has been working great for several days before I put up the second replica while I was working out some hardware issues (in fact the intent of the second replica is to exercise that hardware to figure out what is causing a drive fault)...
I start a replica with these commands:
Code:
zfs snapshot -r $volume@repl-init.$$
zfs send -R $volume@repl-init.$$ | ssh $target zfs receive -F $destvol
Then every 5 minutes, run the mirror script (inside a locker wrapper "just in case" a replica takes too long; they're running about 20 sec typically, up to a couple minutes occasionally):
Code:
touch /vol/.mirror-stamp
zfs destroy -r vol@repl.1
zfs rename -r vol@repl vol@repl.1
zfs snapshot -r vol@repl
zfs send -R -i repl.1 vol@repl | ssh nas03a zfs receive -F vol
These are still a work in progress and the latter hasn't been parameterized yet, and in fact the initialize script hadn't been either, which led to the problem: I initialized the third server as another replica, and in doing so, rotated the repl snapshot so the incremental to the first replica could no longer be done. I grumbled at myself and then re-initialized the first replica, however now when I run the replication, I get these errors on the replica server:
Code:
Feb 22 12:15:21 nas03a mountd[1059]: can't delete exports for /vol/data: Invalid argument
Feb 22 12:15:21 nas03a mountd[1059]: can't delete exports for /vol/vm-images: Invalid argument
Feb 22 12:15:21 nas03a mountd[1059]: bad exports list line /vol/data/home
... repeat for all the filesystems in vol/data
Feb 22 12:15:21 nas03a mountd[1059]: can't change attributes for /vol/vm-images: Invalid radix node head, rn: 0 0xfffff8014d779e00
Feb 22 12:15:21 nas03a mountd[1059]: bad exports list line /vol/vm-images -network 10.1.1.0/24 -maproot
Feb 22 12:15:21 nas03a mountd[1059]: bad exports list line /vol/vm-images/mail.batie.org
Feb 22 12:15:21 nas03a mountd[1059]: bad exports list line /vol/vm-images/ns1
... repeat for the filesystems in vol/vm-images
/etc/exports is a 0 length file (it's not used, but the system kept complaining if it wasn't there)
/etc/zfs/exports is, for all the filesystems on vol:
Code:
/vol/data -network 10.1.1.0/24 -maproot=0
/vol/data/home -network 10.1.1.0/24 -maproot=0
...
It hasn't changed since I got the system setup in the first place and the replication has been working great for several days before I put up the second replica while I was working out some hardware issues (in fact the intent of the second replica is to exercise that hardware to figure out what is causing a drive fault)...