I've been having some issues with ZFS when disks are inserted in a thunderbolt DAS enclosure - a Highpoint Rocketstor 6324LS. My computer is a MacMini 5,1. The Pool is set up as a 3 way mirror, though only two of the disks are currently in use. It works fine (ableit slowly, as I'm stuck with USB2 on the Mini) with the disks in a USB dock. However as soon as I start to move any (or all) of the disks into the thunderbolt enclosure I have issues with io hangs on the pool every (what looks to be) 30 seconds. The pool is currently resilvering so I'd expect to see constant io, which I do on USB, but not with any (or all) disks in the thunderbolt enclosure. See the iostat output below:
- Code: Select all
Sudo zpool iostat 5
capacity operations bandwidth
pool alloc free read write read write
---------- ----- ----- ----- ----- ----- -----
PlayPool 423G 508G 205 44 19.7M 216K
PlayPool 423G 508G 2 0 156K 0
PlayPool 423G 508G 2 0 6.89K 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 14.4K 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 91.0K 1.59K
PlayPool 423G 508G 46 118 203K 481K
PlayPool 423G 508G 60 112 222K 408K
PlayPool 423G 508G 43 103 167K 421K
PlayPool 423G 508G 49 103 186K 411K
PlayPool 423G 508G 19 40 71.0K 149K
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 24 69 85.3K 268K
PlayPool 423G 508G 98 200 376K 791K
PlayPool 423G 508G 10 21 38.2K 76.1K
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 58 121 218K 479K
PlayPool 423G 508G 103 204 399K 807K
PlayPool 423G 508G 102 217 394K 856K
PlayPool 423G 508G 110 225 674K 1.36M
PlayPool 423G 508G 106 171 272K 463K
PlayPool 423G 508G 194 236 1.69M 1.10M
PlayPool 423G 508G 199 57 16.3M 278K
PlayPool 423G 508G 211 34 25.5M 161K
PlayPool 423G 508G 218 29 26.8M 142K
PlayPool 423G 508G 92 1 11.4M 8.80K
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 41 23 4.86M 116K
PlayPool 423G 508G 129 0 16.2M 3.30K
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 8 20 781K 97.1K
PlayPool 423G 508G 218 4 27.4M 20.8K
PlayPool 423G 508G 100 25 12.1M 124K
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 26 24 3.01M 118K
PlayPool 423G 508G 16 0 2.12M 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 0 0 0 0
PlayPool 423G 508G 107 24 13.2M 117K
PlayPool 423G 508G 0 0 0 0
If I've also got another disk in the DAS, formatted as a standard HFS+ volume. If I try and do any significant operations on the HFS volume (rsync a directory, or copy a single large file onto the disk) then the HFS volume will be ejected, with subsequent warnings from OS X. I'd put it down to a problem with the DAS, however if I try and do the same sort of operations with _only_ HFS disks in DAS I haven't been able to get it to fail (though I haven't had the time to really push it yet).
I'm not sure where to go as far as investigating this further - are there specific log files I should be looking at? I've been checking /var/log/system.log, but can't see anything related to the ~30sec timeouts.