RE: "df -h" shows less snapshots than "snap delta"

From: George T Chen (gtchen@yahoo-inc.com)
Date: Thu May 22 2008 - 15:48:04 EDT

  • Next message: Villabroza, Gerald: "Re: NetApp & Leopard"

    What version of OnTap are you running? I have found that versions up to
    at least 7.1 to be inaccurate in their delta measurement.

    > -----Original Message-----
    > From: owner-toasters@mathworks.com
    [mailto:owner-toasters@mathworks.com]
    > On Behalf Of Stephen C. Losen
    > Sent: Thursday, May 22, 2008 4:30 AM
    > To: Borzenkov, Andrey
    > Cc: toasters@mathworks.com
    > Subject: Re: "df -h" shows less snapshots than "snap delta"
    >
    > > I am rather puzzled.
    > >
    > > cn1:~ # rsh filer df -h vol1
    > > Filesystem total used avail capacity Mounted on
    > > /vol/vol1/ 2448GB 776GB 1671GB 32% /vol/vol1/
    > > snap reserve 612GB 158GB 453GB 26% /vol/vol1/..
    > >
    > > So according to this we have 156GB worth of snapshots. So far so
    good.
    > >
    > > cn1:~ # rsh filer snap delta vol1
    > >
    > > Volume vol1
    > > working...
    > >
    > > ...
    > >
    > > Summary...
    > >
    > > From Snapshot To KB changed Time Rate
    > (KB/hour)
    > > --------------- -------------------- ----------- ------------
    ----------
    > ----
    > > Oldest_snap Active File System 265800060 14d 09:08
    770120.583
    > >
    > > Oops. I would expect df -h to show *more* than snap delta (due to
    the
    > fact that we have more intermediate snapshots); but how comes that -
    > apparently having 260GB worth of snapshot data - I only see 158GB in
    space
    > accounting?
    > >
    > > I am likely missing something obvious. Hmm ... is it possible that
    "snap
    > delta" accounts for both changed *and* new data while "df -h" accounts
    for
    > *changed* data only? Looks plausible...
    >
    > Snapshot accounting can be very tricky I think. Consider this
    scenario.
    >
    > 1) take snapshot #1 of a volume
    >
    > 2) create a new and very large file
    >
    > 3) take snapshot #2 of volume
    >
    > 4) delete the very large file
    >
    > 5) take snapshot #3
    >
    > At this point the "delta" between snapshots 1 and 3 and the live
    volume
    > will be very small. But the "delta" between snapshot 2 and any other
    > snapshot (or the live volume) will be large. This is because snapshot
    2
    > (and only snapshot 2) still contains the large file that was deleted.
    >
    > So you can see that the delta between the live volume and each of
    > its snapshots does not necessarily increase with the age of the
    snapshot.
    >
    >
    > Steve Losen scl@virginia.edu phone: 434-924-0640
    >
    > University of Virginia ITC Unix Support
    >
    >



    This archive was generated by hypermail 2b29 : Thu May 22 2008 - 16:45:56 EDT