My photo-taking workflow while on vacation usually involves taking a lot of photos daily, dumping them to a laptop, processing, then backing them up once I have returned home.
Previously, I accomplished this manually using BeyondCompare for Windows, as that would run on Windows Home Server.
Since moving to ZFS-based storage, however, this is no longer an option as BeyondCompare only has a Linux client (nothing for Unix/BSD).
There are other ways to get around this:
- SSHFS and Meld – Complicated, somewhat bloated, but great BeyondCompare alternative
- *Commander Utilities – Midnight Commander derivatives can accomplish similar tasks using the ctrl+x,d shortcut
- Rsync – typically installed by default, easy to script
I chose Rsync as I wanted something more automated, but I do find myself using Midnight Commander from time-to-time to simply “get things done” when syncing files other than my images.
Here’s how I did it:
rsync -a -e ssh /volumes/PICTURES/ 'username@mymac:/Volumes/BIGRAID/'
Let’s break this down into smaller pieces:
rsync – this is the command that will do our heavy lifting and file comparison
-a – archive mode
-e – specify an RSH replacement
ssh – use SSH
/volumes/PICTURES/ – this specifies the “Volumes” folder on my Mac, and the “PICTURES” drive within it. Replace this with the location of your items to backup
‘ – note the use of single quotes here. We’re using these in case there are spaces in the folder names, and we could have done the same above.
username@mymac – We’re logging on to the host “mymac” with the username “username”. You’ll probably want to change these. I use a hostname here, but you could just as easily use an IP address if you use static IP addresses.
:/volumes/BIGRAID/ – the colon denotes a subfolder on the server we are backing up to, and /volumes/BIGRAID in this case refers to a ZFS pool called “BIGRAID”.
Do you have a similar backup strategy for BSD/Unix targets that you would like to share?
If you’ve been running snapshots for a while and have already backed them up, you might occasional need to delete all zfs snapshots for your pool.
Typically, you’d do this as part of your backup script, assuming that they have been written correctly.
First, to find the used snapshot space, run this command:
zfs list -o space
This will give you a detailed readout of your pools and snapshot space used.
Here’s my script to wipe ZFS snap shots, but I am certainly open to suggestion:
zfs list -H -o name -t snapshot | xargs -n1 zfs destroy
Again, caution is needed as this will remove ALL SNAPS from your pools.
I’ve decided to replace the Windows Home Server Vail server with something capable of handling newer builds of ZFS and the inherent deduploication.
Here’s a quick kit list and build diary I’ll try to keep up-to-date as I go along.
- Dell Perc6i – this is essentially a port multiplier. I scored it from eBay on the cheap, though it was delivered from Israel, took awhile, and had neither cables nor mounting bracket.
- OCZ RevoDrive 120GB – Though the RAID controller on this card is not supported in Linux/Solaris, the drives show up as two separate devices as long as you make sure to put it in the right PCIe slot. That means it’s perfect for both ZIL (log) and L2ARC (cache).
- 2x Intel 80GB X25-M SSDs – these will house the virtual machine files to be deduped. Very reliable drives, and though they might not be the fastest in terms of writes, the speeds are relatively constant which is quite handy compared to solutions that attempt compression like SandForce controllers. ZFS will take care of that, thanks.
- (IN TRANSIT) 2x Dual Port 1gbit Intel PCIe NICs – I’ll use these for the direct connection to the virtual machine host. Currently one link is used, but when reading from the SSD drives the line is saturated.
- (IN TRANSIT) 32 Pin SAS Controller To 4x SATA HDD Serial Cable Cord – This is needed to plug in 8 drives to the LSI controller.
- 5x 1.5TB Seagate hard drives – These will be the bread-and-butter storage running in RAID-Z2 (similar to RAID 6).
- 3x 3TB Seagate hard drives – These might simply be a large headache, but the plan was to have an extra 3TB RAID-Z2 for backups in another machine. Unfortunately there seem to be issues with drives that are 4k presenting themselves as 512b. I may be able to get around this by hacking or waiting as they become more popular. For now 2 of them are in software RAID1 on a Windows 7 host, and the other remains in the external USB 3 case and is used as a backup drive.
- NetGear GS108T Switch – A cheap VLAN-capable switch should I decide to use more than 2 bonded ports (I doubt it), currently running the lab.
EMC Corp. has acquired a startup still in stealth called Pi Corp. in
an all-cash transaction for an undisclosed amount.
Pi Corp. founder and CEO Paul Maritz will join EMC as president and
general manager of EMC’s newly created best cloud storage infrastructure and
Services Division, reporting directly to EMC CEO Joe Tucci. The new
division will also include the EMC Fortress SaaS infrastructure, the
Mozy online backup service and “other upcoming EMC cloud
infrastructure systems and software offerings under development.”
This is interesting to me as I see it as the “new next big thing”.
Currently the only people really loud in this space are Google and Amazon.
I highly doubt we will see much come out of either party until the very soonest Q2 2008 (Mozy update), but expect a product by year-end.