NETGEAR is aware of a growing number of phone and online scams. To learn how to stay safe click here.

Forum Discussion

NessD's avatar
NessD
Follower
Apr 12, 2013

Why UPS for Spin-Down?

Hi!
Just a quick question I couldn't find the answer for. Why is it recommended to use an UPS, when you turn on HDD spin-down? How can data get lost? I'm doing an incremental backup once or twice a day. Is there any serious risk of losing all of it, when turning on spin-down and having a power-failure?

Thanks
NessD

8 Replies

  • Personally I recommend a UPS whether HDD Spin-down is enabled or not.

    When data is written to the disks, it is queued up (and not written immediately). So if the power fails, the stuff waiting in the queue never gets written. When you use RAID, every thing needs to be written to at least 2 drives. Where things get particularly dodgy is when one of the drives gets written to, but the other[s] don't. That can put the RAID array out of sync, and do damage to the file system integrity. This might do minor damages, but if you are unlucky it can destroy everything - it all depends on what didn't make it to the disk, and what the NAS tries to do to repair the damage.

    If the disks are spun down when you update data, then they need to be spun up first before anything can be written. While this is happening the queues can get pretty big, so there is more risk of corruption if the power happens to be cut during this time.

    If you look through the posts here, you will find several from folks who weren't using a UPS and who lost all their data when the power failed. Its not something that happens every time, but it certainly is a real risk.
  • Well I disagree... First, there is a chance data is lost or corrupted when you have a lower outage. However....
    - you cannot lose all of your data in this way - even if for some reason the RAID volume becomes unmountable the data is still there and can be recovered.
    - ext3/4 has ordered journalling by default, and I'm sure that this is the default for ReadyNAS. In this case I understand that you could get some corrupt data if it is being overwritten at the time, but that the journal replay should bring it back to a stable state.

    Otherwise, what's the point in journaling. IMHO a UPS is handy, but only if you perform an ordered shutdown when it is enabled. In any case, you can of course recover the data from your backup, right :-)

    P.S. And again IMHO I have yet to see a huge benefit of spin-down. Sure you can save a bit of noise and watts, but you do place most strain on the PSU and disks when spinning back up.
  • ihartley wrote:
    ....you cannot lose all of your data in this way - even if for some reason the RAID volume becomes unmountable the data is still there and can be recovered...
    Two of many posts:
    viewtopic.php?t=70092&p=388851
    viewtopic.php?f=64&t=57392

    It's not clear if these two got their data back, perhaps they did. Even if it is potentially recoverable, most users don't have the skills. Often a recovery attempt creates more damage.

    Independent of spin-down settings, I still recommend UPS protection for ReadyNAS.
  • StephenB : respect! I don't disagree with having a UPS as a recommendation (even though I don't). But I get the distinct feeling that Netgear/ReadyNAS problems are all too easily blamed in drives. Drive that - if you forgive me - have a common, standardised interface and work in millions of other systems.

    I do have a number of issues with the posts you reference:
    - "RAID is not sufficient to protect your data from loss" - no but it is sufficient - and designed - to protect your data from single (or multiple depending on config) HDD failure
    - "surge protection" - I would suggest that the PSU would handle this quite well, and most ethernet/cable feeds are already protected.
    - "disks failing at the same time" - statistically, this is as likely to happen as winning the lottery. But people do win it!!

    I think you hit the nail on the head though. People plug/unplug drives, and in doing so accidentally wipe data. I would recommend that the first thing to do when you suspect whole data loss is remove drives and not re-plug them. It seems Netgear have decided to treat any new drive inserted as a new drive. Why - I have no idea.....
  • We agree more than we disagree. One reason I recommend a UPS is that I lost a Duo v1 some years ago w/o one. Perhaps now I could have recovered it; one benefit of being active in the forum is that you learn stuff from other posters.

    However, since I put the UPS in place I have always had controlled shutdowns (or recovered power before the battery discharged). I've had no issues with file loss with any of my ReadyNAS since then (now up to 4, though the duo v1 will be retired soon). Since a UPS costs about the same as a disk drive, it seems like good insurance to me.

    On RAID, I guess I'd say it is more about maintaining availability than protection. If you can afford the drives, I think you get better protection with multiple copies on different jbod volumes. Certainly recovery is simpler (unless we are talking RAID-1).

    On disks failing at the same time: there are studies suggesting that this is not just a lottery - that in fact disk failures are correlated in RAID arrays. Multiple failures happen much more often than the statistics predict. There are some reasons why this seems credible. People tend to purchase identical drives in batches - particularly the large data centers that were studied in this research. Even home NAS users also frequently buy new drives when setting up their NAS. The drives are likely to be in the same manufacturing lot, and shipped together. Also, drives in the same RAID chassis are under near-identical loads, experience near-identical temperature/vibration, are fed by the same power supply, etc. Identical drives, identical handling, identical environment -> failures happen close together.

    EDIT- Just wanted to add that I have two friends who lost almost all their electronics due to a power surge (nearby lightening strikes in both cases). There was clear evidence that the surge went through their data cables as well as their power.
  • Get a UPS, end of story. To save $50 and not have a UPS is a mistake. Look at the APC ES550 which includes the USB monitoring cable. Your NAS and its data will thank you.

    And, BTW, if you're very concerned about surges, the surge protection in a $50 UPS is fairly basic. I recommend you invest in a proper catastrophic surge protector. i.e. Panamax Max series.
  • Aha, you bring up an interesting one! I suppose it all depends where you live; e.g. here in Blighty the good 'ol IEE and BT between them have done an excellent job of protecting surges to both the supply and data comms. I guess if you're in India you'll need your own UPS connected to a generator :-)

    But since almost all HDDs now auto-retract on power loss, I still assert that at worst your disks can become inconsistent and/or lose or corrupt data if writing when power loss. I would rather suggest that the power loss is the first time in a long time that the RAID array has been rebooted, and that the problems have been building for some time.

    I agree a UPS is a no-brainer. But not having one does not cause complete data loss.
    P.S. StephenB: Sorry, yes you are right - correlation between failures in drive batches is well documented. Outside of this I'm not convinced. Or to put it another way "What's the difference between an Actuary and a Mafia Boss?? The Actuary can predict when you'll die, the Mafia Boss can predict how..!" :-)
  • ihartley wrote:
    ..."What's the difference between an Actuary and a Mafia Boss?? The Actuary can predict when you'll die, the Mafia Boss can predict how..!"...
    ROTFL :D

NETGEAR Academy

Boost your skills with the Netgear Academy - Get trained, certified and stay ahead with the latest Netgear technology! 

Join Us!

ProSupport for Business

Comprehensive support plans for maximum network uptime and business peace of mind.

 

Learn More