Useful stuff for your blog-reading pleasure.

All | General

Tuesday August 12, 2008

ZFS saved my data. Right now.

Kid with a swimming ring. As you know, I have a server at home that I use for storing all my photos, music, backups and more using the Solaris ZFS filesystem. You could say that I store my life on my server.

For storage, I use Western Digital's MyBook Essential Edition USB drives because they are the cheapest ones I could find from a well-known brand. The packaging says "Put your life on it!". How fitting.

Last week, I had a team meeting and a colleague introduced us to some performance tuning techiques. When we started playing with iostat(1M), I logged into my server to do some stress tests. That was when my server said something like this:

constant@condorito:~$ zpool status

(data from other pools omitted)

  pool: santiago
 state: DEGRADED
status: One or more devices has experienced an unrecoverable error.  An
	attempt was made to correct the error.  Applications are unaffected.
action: Determine if the device needs to be replaced, and clear the errors
	using 'zpool clear' or replace the device with 'zpool replace'.
   see: http://www.sun.com/msg/ZFS-8000-9P
 scrub: scrub completed after 16h28m with 0 errors on Fri Aug  8 11:19:37 2008
config:

	NAME         STATE     READ WRITE CKSUM
	santiago     DEGRADED     0     0     0
	  mirror     DEGRADED     0     0     0
	    c10t0d0  DEGRADED     0     0   135  too many errors
	    c9t0d0   DEGRADED     0     0    20  too many errors
	  mirror     ONLINE       0     0     0
	    c8t0d0   ONLINE       0     0     0
	    c7t0d0   ONLINE       0     0     0

errors: No known data errors

This tells us 3 important things:

Two of my disks (c10t0d0 and c9t0d0) are happily giving me garbage back instead of my data. Without knowing it.
Thanks to ZFS' checksumming, we can detect this, even though the drive thinks everything is ok.
No other storage device, RAID array, NAS or file system I know of can do this. Not even the increasingly hyped (and admittedly cool-looking) Drobo [1].
Because both drives are configured as a mirror, bad data from one device can be corrected by reading good data from the other device. This is the "applications are unaffected" and "no known data errors" part.
Again, it's the checksums that enable ZFS to distinguish good data blocks from bad ones, and therefore enabling self-healing while the system is reading stuff from disk.
As a result, even though both disks are not functioning properly, my data is still safe, because (luckily, albeit with millions of blocks per disk, statistics is on my side here) the erroneous blocks don't overlap in terms of what pieces of data they store.
Again, no other storage technology can do this. RAID arrays only kick in when the disk drives as a whole are unacessible or when a drive diagnoses itself to be broken. They do nothing against silent data corruption, which is what we see here and what all people on this planet that don't use ZFS (yet) can't see (yet). Until it's too late.
Data hygiene is a good thing. Do a "zpool scrub <poolname>" once in a while. Use cron(1M) to do this, for example every other week for all pools.

Over the weekend, I ordered myself a new disk (sheesh, they dropped EUR 5 in price already after just 5 days...) and after a "zpool replace santiago c10t0d0 c11t0d0" on monday, my pool started resilvering:

constant@condorito:~$ zpool status

(data from other pools omitted)

  pool: santiago
 state: DEGRADED
status: One or more devices has experienced an unrecoverable error.  An
        attempt was made to correct the error.  Applications are unaffected.
action: Determine if the device needs to be replaced, and clear the errors
        using 'zpool clear' or replace the device with 'zpool replace'.
   see: http://www.sun.com/msg/ZFS-8000-9P
 scrub: resilver in progress for 1h13m, 6.23% done, 18h23m to go
config:

        NAME           STATE     READ WRITE CKSUM
        santiago       DEGRADED     0     0     0
          mirror       DEGRADED     0     0     0
            replacing  DEGRADED     0     0     0
              c10t0d0  DEGRADED     0     0   135  too many errors
              c11t0d0  ONLINE       0     0     0
            c9t0d0     DEGRADED     0     0    20  too many errors
          mirror       ONLINE       0     0     0
            c8t0d0     ONLINE       0     0     0
            c7t0d0     ONLINE       0     0     0

errors: No known data errors

The next step for me is to send the c10t0d0 drive back and ask for a replacement under warranty (it's only a couple of months old). After receiving c10's replacement, I'll consider sending in c9 for replacement (depending on how the next scrub goes).

Which makes me wonder: How will drive manufacturers react to a new wave of warranty cases based on drive errors that were not easily detectable before?

[1] To the guys at Drobo: Of course you're invited to implement ZFS into the next revision of your products. It's open source. In fact, Drobo and ZFS would make a perfect team!

"ZFS saved my data. Right now." has been brought to you by Constantin's Blooog.
This entry was created on 2008-08-12 06:44:22.0 PST and is associated with the following tags:
corruption data drobo integrity opensolaris solaris storage zfs
You're welcome to use this Permalink , add a comment below or send your feedback to constantin at sun dot com.
Comments [5]

Thursday September 06, 2007

7 Easy Tips for ZFS Starters

So you're now curious about ZFS. Maybe you read Jonathan's latest blog entry on ZFS or you've followed some other buzz on the Solaris ZFS file system or maybe you saw a friend using it. Now it's time for you to try it out yourself. It's easy and here are seven tips to get you started quickly and effortlessly:

1. Check out what Solaris ZFS can do for you

First, try to compose yourself a picture of what the Solaris ZFS filesystem is, what features it has and how it can work to your advantage. Check out the CSI:Munich video for a fun demo on how Solaris ZFS can turn 12 cheap USB memory sticks into highly available, enterprise-class, robust storage. Of course, what works with USB sticks also works with your own harddisks or any other storage device. Also, there are great ZFS screencasts that show you some more powerful features in an easy to follow way. Finally, there's a nice writeup on "What is ZFS?" at the OpenSolaris ZFS Community's homepage.

2. Read some (easy) documentation

It's easy to configure Solaris ZFS. Really. You just need to know two commands: zpool (1M) and zfs (1M). That's it. So, get your hands onto a Solaris system (or download and install it for free) and take a look at those manpages. If you still want more, then there's of course the ZFS Administration Guide with detailed planning, configuration and troubleshooting steps. If you want to learn even more, check out the OpenSolaris ZFS Community Links page. German-speaking readers are invited to read my german white paper on ZFS or listen to episode #006 of the POFACS podcast.

3. Dive into the pool

Solaris ZFS manages your storage devices in pools. Pools are a convenient way of abstracting storage hardware and turning it into a repository of blocks to store your data in. Each pool takes a number of devices and applies an availability scheme (or none) to it. Pools can then be easily expanded by adding more disks to them. Use pools to manage your hardware and its availability properties. You could create a mirrored pool for data that should be protected against disk failure and that needs fast access to hardware. Then, you could add another pool using RAID-Z (which is similar, but better than RAID-5) for data that needs to be protected but where performance is not the first priority. For scratch, test or demo data, a pool without any RAID scheme is ok, too. Pools are easily created:

zpool create mypool mirror c0d0 c1d0

Will create a mirror out of the two disk devices c0d0 and c1d0. Similarly, you can easily create a RAID-Z pool by saying:

zpool create mypool raidz c0d0 c1d0 c2d0

The easiest way to turn a disk into a pool is:

zpool create mypool c0d0

It's that easy. All the complexity of finding, sanity-checking, labeling, formatting and managing disks is hidden behind this simple command.

If you don't have any spare disks to try this out with, then you can just create yourself some files, then use them as if they were block devices:

# mkfile 128m /export/stuff/disk1 # mkfile 128m /export/stuff/disk2 # zpool create testpool mirror /export/stuff/disk1 /export/stuff/disk2 # zpool status testpool pool: testpool state: ONLINE scrub: none requested config: NAME STATE READ WRITE CKSUM testpool ONLINE 0 0 0 mirror ONLINE 0 0 0 /export/stuff/disk1 ONLINE 0 0 0 /export/stuff/disk2 ONLINE 0 0 0 errors: No known data errors

The cool thing about this procedure is that you can create as many virtual disks as you like and then test ZFS's features such as data integrity, self-healing, hot spares, RAID-Z and RAID-Z2 etc. without having to find any free disks.

When creating a pool for production data, think about redundancy. There are three basic properties to storage: availability, performance and space. And it's a good idea to prioritize them in that order: Make sure you have redundancy (mirroring, RAID-Z, RAID-Z2) so ZFS can self-heal data when stuff goes wrong at the hardware level. Then decide how much performance you want. Generally, mirroring is faster and more flexible than RAID-Z/Z2, especially if the pool is degraded and ZFS needs to reconstruct data. Space is the cheapest of all three, so don't be greedy and try to give priority to the other two. Richard Elling has some great recommendations on RAID, space and MTTDL. Roch has also posted a great article on mirroring vs. RAID-Z.

4. The power to give

Once you have set up your basic pool, you can already access your new ZFS file system: Your pool has been automatically mounted for you in the root directory. If you followed the examples above, then you can just cd to /mypool and start using ZFS!

But there's more: Creating additional ZFS file systems that use your pool's resources is very easy, just say something like:

zfs create mypool/home zfs create mypool/home/johndoe zfs create mypool/home/janedoe

Each of these commands only takes seconds to complete and every time you will get a full new file system, already set up and mounted for you to start using it immediately. Notice that you can manage your ZFS filesystems hierarchically as seen above. Use pools to manage storage properties at the hardware level, use filesystems to present storage to your users and applications. Filesystems have properties (compression, quotas, reservations, etc.) that you can easily administer using zfs set and that are inherited across the hierarchy. Check out Chris Gerhard's blog on more thoughts about file system organization.

5. Snapshot early, snapshot often

ZFS snapshots are quick, easy and cheap. Much cheaper than the horrible experience when you realize that you just deleted a very important file that hasn't been backed up yet! So, use snapshots whenever you can. If you think about whether to snapshot or not, just do it. I recently spent only about $220 on two 320 GB USB disks for my home server to expand my pool with. At these prices, the time you spend thinking about whether to snapshot or not may be more worth than just buying more disk.

Again, Chris has some wisdom on this topic in his ZFS snapshot massacre blog entry. He once had over 60000 snapshots and he's snapshotting filesystems by the minute! Since snapshots in ZFS “just work” and since they only take up the space that actually changes between snapshots, there's really no reason to not doing snapshots all the time. Maybe once per minute is a little bit exaggerated, but once a week, once per day or once an hour per active filesystem is definitely good advice.

Instead of time based snapshotting, Chris came up with the idea to snapshot a file system shared with Samba whenever the Samba user logs in!

6. See the Synergy

ZFS by itself is very powerful. But the full beauty of it can be unleashed by combining ZFS with other great Solaris 10 features. Here are some examples:

Tim Foster has written a great SMF service that will snapshot your ZFS filesystems on a regular basis. It's fully automatic, configurable and integrated with SMF in a beautiful way.
ZFS can create block devices, too. They are called zvols. Since Nevada build 54, they are fully integrated into the Solaris iSCSI infrastructure. See Ben Rockwood's blog entry on the beauty of iSCSI with ZFS.
A couple of people are now elevating this concept even further: Take two Thumpers, create big zvols inside them, export them through iSCSI and mirror over them with ZFS on a server. You'll get a huge, distributed storage subsystem that can be easily exported and imported on a regular network. A poor man's SAN and a powerful shared storage for future HA clusters thanks to ZFS, iSCSI and Thumper! Jörg Möllenkamp is taking this concept a bit further by thinking about ZFS, iSCSI, Thumper and SAM-FS.
Check out some cool Sun StorageTek Availability Suite and ZFS demos here.
ZFS and boot support is still in the works, but if you're brave, you can try it out with the newer Solaris Nevada distributions on x64 systems. Think about the possibilities together with Solaris Live Upgrade! Create a new boot environment in seconds while not needing to find or dedicate a new partition, thanks to snapshots, while saving most of the needed disk space!

And that's only the beginning. As ZFS becomes more and more adopted, we'll see many more creative uses of ZFS with other Solaris 10 technologies and other OSes.

7. Beam me up, ZFS!

One of the most amazing features of ZFS is zfs send/receive. zfs send will turn a ZFS filesystem into a bitstream that you can save to a file, pipe through bzip2 for compression or send through ssh to a distant server for archiving or for remote replication through the corresponding zfs receive command. It also supports incremental sending and receiving out of subsequent snapshots through the -i modifier.

This is a powerful feature with a lot of uses:

Create your Solaris zone as a ZFS filesystem, complete with applications, configuration, automation scripts, users etc., zfs send | bzip2 >zone_archive.zfs.bz2 it for later use. Then, unpack and create hundreds of cloned zones out of this master copy.
Easily migrate ZFS filesystems between pools on the same machine or on distant machines (through ssh) with zfs send/receive.
Create a crontab entry that takes a snapshot every minute, then zfs send -i it over ssh to a second machine where it is piped into zfs receive. Tadah! You'll get free, finely-grained, online remote replication of your precious data.
Easily create efficient full or incremental backups of home directories (each in their own ZFS filesystems) through ZFS send. Again, you can compress them and treat them like you would, say, treat a tar archive.

See? It is easy, isn't it? I hope this guide helps you find your way around the world of ZFS. If you want more, drop by the OpenSolaris ZFS Community, we have a mailing list/forum where bright and friendly people hang out that will be glad to help you.

"7 Easy Tips for ZFS Starters" has been brought to you by Constantin's Blooog.
This entry was created on 2007-09-06 11:20:15.0 PST and is associated with the following tags:
adoption community data filesystem free howto innovation introduction open opensolaris opensource software solaris storage tips unix zfs
You're welcome to use this Permalink , add a comment below or send your feedback to constantin at sun dot com.