• Home
  • About

Saved searches

  • Remove
  • Verified account @
Suggested users
  • Verified account @
  • Verified account @
  • Language: English
    • Bahasa Indonesia
    • Bahasa Melayu
    • Català
    • Čeština
    • Dansk
    • Deutsch
    • English UK
    • Español
    • Filipino
    • Français
    • Hrvatski
    • Italiano
    • Magyar
    • Nederlands
    • Norsk
    • Polski
    • Português
    • Română
    • Slovenčina
    • Suomi
    • Svenska
    • Tiếng Việt
    • Türkçe
    • Ελληνικά
    • Български език
    • Русский
    • Српски
    • Українська мова
    • עִבְרִית
    • العربية
    • فارسی
    • मराठी
    • हिन्दी
    • বাংলা
    • ગુજરાતી
    • தமிழ்
    • ಕನ್ನಡ
    • ภาษาไทย
    • 한국어
    • 日本語
    • 简体中文
    • 繁體中文
  • Have an account? Log in
    Have an account?
    · Forgot password?

    New to Twitter?
    Sign up
By using Twitter’s services you agree to our Cookie Use and Data Transfer outside the EU. We and our partners operate globally and use cookies, including for analytics, personalisation, and ads.
cmuratori's profile
Casey Muratori
Casey Muratori
Casey Muratori
@cmuratori

Casey Muratori

@cmuratori

Programmer at http://mollyrocket.com  on @molly1935 (https://molly1935.com ) and host of @handmade_hero (https://handmadehero.org )

Joined March 2009
  • © 2016 Twitter
  • About
  • Help
  • Terms
  • Privacy
  • Cookies
  • Ads info
Dismiss
Previous
Next

Go to a person's profile

Saved searches

  • Remove
  • Verified account @
Suggested users
  • Verified account @
  • Verified account @

Retweet this to your followers?

Optional comment for Retweet
 
 

Saved searches

  • Remove
  • Verified account @
Suggested users
  • Verified account @
  • Verified account @
140

Are you sure you want to delete this Tweet?

Promote this Tweet

Block

  • Add a location to your Tweets

    When you tweet with a location, Twitter stores that location. You can switch location on/off before each Tweet and always have the option to delete your location history. Learn more

    Profile summary

    Your lists

    Create a new list


    Under 100 characters, optional

    Privacy

    Your reply includes the people in this conversation up to this point. Learn more

    Copy link to Tweet

    Embed this Tweet

    Embed this Video

    Add this Tweet to your website by copying the code below. Learn more

    Add this video to your website by copying the code below. Learn more

    Hmm, there was a problem reaching the server.

    Preview

    Log in to Twitter

    · Forgot password?
    Don't have an account? Sign up »

    Sign up for Twitter

    Not on Twitter? Sign up, tune into the things you care about, and get updates as they happen.

    Sign up
    Have an account? Log in »

    Two-way (sending and receiving) short codes:

    Country Code For customers of
    United States 40404 (any)
    Canada 21212 (any)
    United Kingdom 86444 Vodafone, Orange, 3, O2
    Brazil 40404 Nextel, TIM
    Haiti 40404 Digicel, Voila
    Ireland 51210 Vodafone, O2
    India 53000 Bharti Airtel, Videocon, Reliance
    Indonesia 89887 AXIS, 3, Telkomsel, Indosat, XL Axiata
    Italy 4880804 Wind
    3424486444 Vodafone
    » See SMS short codes for other countries

    Confirmation

     

    Buy Now

    Hmm... Something went wrong. Please try again.

    Previous Tweet
    1. Casey Muratori ‏@cmuratori 17 Oct 2013

      Does anyone out there understand the decision to store ZFS uberblocks in a _continuous_ array?

      0 retweets 0 likes
    2. Casey Muratori ‏@cmuratori 17 Oct 2013

      @cmuratori It seems to me that is basically just asking to have that array (which will all be on one SSD page) get wiped out together.

      0 retweets 0 likes
    3. Casey Muratori ‏@cmuratori 17 Oct 2013

      @cmuratori Why wouldn't you store uberblocks _strided_ across the drive instead?

      0 retweets 0 likes
    4. Casey Muratori ‏@cmuratori 17 Oct 2013

      @cmuratori The ZFS people seem smart so I'm wondering if I'm just not getting why it's not a problem for the uberblocks to be contiguous.

      0 retweets 0 likes
    5. Won Chun ‏@won3d 17 Oct 2013

      @cmuratori I'm guessing uberblocks are replicated. And they are probably not striped because syncs across disks aren't transactional.

      0 retweets 0 likes
    6. Sean Barrett ‏@nothings 17 Oct 2013

      @won3d @cmuratori (2) from same link: only *1* uberblock is actually the current valid root, you don't care what's in the other 127 anyway

      0 retweets 0 likes
    7. Casey Muratori ‏@cmuratori 17 Oct 2013

      @nothings @won3d Yes but that's not relevant here. Since the 128 are stored together, if that sector/pagegoes they will all go.

      0 retweets 0 likes
    8. Sean Barrett ‏@nothings 17 Oct 2013

      @cmuratori @won3d But only one of them is the current root. The other 127 are stale roots, so you don't care if you lose them.

      0 retweets 0 likes
    9. Casey Muratori ‏@cmuratori 17 Oct 2013

      @nothings @won3d No, that is incorrect. They are all "valid roots" due to COW, so in case the latest one had a bad write, you are OK.

      0 retweets 0 likes
    10. Sean Barrett ‏@nothings 17 Oct 2013

      @cmuratori @won3d Obviously that doesn't work once free space gets recycled. But you're right: http://utcc.utoronto.ca/~cks/space/blog/solaris/ZFSMetadataRecovery …

      0 retweets 0 likes
      Casey Muratori ‏@cmuratori 17 Oct 2013

      @nothings @won3d It actually does work even when free space gets recycled, just not for the entire tree. So you still get _some_ protection.

      2:46 PM - 17 Oct 2013
      0 retweets 0 likes
        1. Casey Muratori ‏@cmuratori 17 Oct 2013

          @nothings @won3d You'll only lose the part that actually got recycled, rather than the whole drive which is what would happen otherwise.

          0 retweets 0 likes
        2. Sean Barrett ‏@nothings 17 Oct 2013

          @cmuratori @won3d It depends on what got recycled. If something high in the tree got recycled... again, see the link I gave for more details

          0 retweets 0 likes
        3. Casey Muratori ‏@cmuratori 17 Oct 2013

          @nothings @won3d It would have to be something high in the tree _on all 127 other uberblocks_.

          0 retweets 0 likes
        4. View other replies
        5. Chris Siebenmann ‏@thatcks 17 Oct 2013

          @cmuratori @nothings @won3d Note: high level metadata is always changing on ZFS because of copy-on-write plus pointers to lower-level data.

          0 retweets 0 likes
        6. Casey Muratori ‏@cmuratori 17 Oct 2013

          @thatcks @nothings @won3d Yes and this worries me a bit, separate from the question I asked earlier about uberblocks in contiguous arrays.

          0 retweets 0 likes
        7. View other replies
        8. Casey Muratori ‏@cmuratori 17 Oct 2013

          @thatcks @nothings @won3d Specifically, because you're always rewriting the uberblocks to the same addresses, you have a "superpage" problem

          0 retweets 0 likes
        9. Casey Muratori ‏@cmuratori 17 Oct 2013

          @thatcks @nothings @won3d Ie., 8K may be the write size for an SSD, but the _clear_ size is actually like half a meg or something.

          0 retweets 0 likes
        10. Casey Muratori ‏@cmuratori 17 Oct 2013

          @thatcks @nothings @won3d So, I think this only maps well to an SSD if the firmware happens to do a full 8K page mapping.

          0 retweets 0 likes
        11. Show more

      Loading seems to be taking a while.

      Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.

        Promoted Tweet

        false

        • © 2016 Twitter
        • About
        • Help
        • Terms
        • Privacy
        • Cookies
        • Ads info