Skip to content
Commits on Jan 19, 2011
  1. @dustin @chiyoung

    Added dbconvert to convert between kvstore implementations.

    dustin committed with chiyoung Jan 18, 2011
    Change-Id: Ia2ce02781bf28a6c74ea573795f2b18cd767524b
    Reviewed-on: http://review.membase.org/4279
    Reviewed-by: Chiyoung Seo <chiyoung.seo@gmail.com>
    Tested-by: Chiyoung Seo <chiyoung.seo@gmail.com>
Commits on Sep 21, 2010
  1. @dustin @chiyoung

    bug 2443 - append+prepend fixes with binary data

    dustin committed with chiyoung Sep 21, 2010
    Change-Id: I121346bfa5e160fe56c7d0a23f11427ee5ea7db5
    Reviewed-on: http://review.northscale.com/2616
    Reviewed-by: Chiyoung Seo <chiyoung.seo@gmail.com>
    Tested-by: Chiyoung Seo <chiyoung.seo@gmail.com>
Commits on Sep 17, 2010
  1. @dustin @chiyoung

    bug 2354 - Always use absolute time.

    dustin committed with chiyoung Sep 16, 2010
    Change-Id: Ic61540264e663739b0d21683497c9b316f0ee177
    Reviewed-on: http://review.northscale.com/2504
    Tested-by: Chiyoung Seo <chiyoung.seo@gmail.com>
    Reviewed-by: Chiyoung Seo <chiyoung.seo@gmail.com>
Commits on Jul 7, 2010
  1. @dustin @seanlynch

    Document Blob.

    dustin committed with seanlynch Jul 3, 2010
    Change-Id: I926e8b8f5ccdeb6a51920468ef15e78ad0590ca6
    Reviewed-on: http://review.northscale.com/1020
    Reviewed-by: Matt Ingenthron <matt@northscale.com>
    Reviewed-by: Sean Lynch <seanl@literati.org>
    Tested-by: Sean Lynch <seanl@literati.org>
Commits on Jul 3, 2010
  1. @dustin

    Merge branch 'z' (stability, performance, size)

    dustin committed Jul 2, 2010
    * z: (31 commits)
      Mod before hash to avoid abs underflow - bug 1681
      Verify we have a tap connection to remove at hangup - bug 1655
      Fixed reconnect handling to properly reset state - bug 1668
      A safer io completion signaler.
      Terminate the tap client on exception.
      Refuse unreasonably large tap requests - bug 1653
      Free leftover message when tearing down tap client - bug 1653
      Throttle backfills while consumers are slow - bug 1635
      Disable dedup support for backfilled tap data - bug 1654
      Striped hashtable visits - bug 1654
      Incrementally hand backfilled data into the tap queue - bug 1654
      Do not send noops if set to logical forever.  Bug 1665.
      Allow a tap ID to be specified in tap.py.
      Disconnect unknown tap clients - bug 1637
      Prevent concurrent backfills for a single tap client - bug 1635
      Only print printable characters in tap.py
      Delete old queue object after completing backfill.
      Report more information on fetch failures.
      Modified tests to handle add and restart + set cases.
      Use static polymorphism to reduce the overhead of StoredItem.
      ...
    
    Conflicts:
    	ep.cc
    	ep.hh
    	ep_engine.h
    	ep_testsuite.c
    	sqlite-pst.cc
    	stored-value.cc
    	stored-value.hh
    
    Change-Id: I3be46e7f2033810185aad0c94f9d7b2c39d0563e
Commits on Jun 25, 2010
  1. @dustin

    New Blob type for storing data values.

    dustin committed Jun 19, 2010
    Change-Id: If2ae62073f94c4b661037c8ab5cf729e697c8b30
    Reviewed-on: http://review.northscale.com/770
    Tested-by: Dustin Sallings <dustin@spy.net>
    Reviewed-by: Sean Lynch <seanl@literati.org>
Commits on Jun 7, 2010
  1. @dustin

    Merge branch 'z'

    dustin committed Jun 7, 2010
    * z:
      Do not index keys on disk.
      Bug 1180: set a max limit for data object size
    
    Conflicts:
    	ep.cc
    	item.hh
    	sqlite-kvstore.cc
    
    Change-Id: Ia827e9816b862186ba32650d6ff573b63a92c7f8
  2. @dustin

    Do not index keys on disk.

    dustin committed Jun 6, 2010
    This, instead keeps a reference to all the k/v pairs in memory along
    with their rowid on disk.  Much smaller and faster persistence.
    
    Change-Id: I9548472e9a0eaaa53d53261293e9016e335b245e
    Reviewed-on: http://review.northscale.com:8080/313
    Tested-by: Trond Norbye <trond.norbye@gmail.com>
    Reviewed-by: Trond Norbye <trond.norbye@gmail.com>
    Reviewed-by: Sean Lynch <seanl@literati.org>
Commits on Jun 3, 2010
  1. @dustin

    Merge branch 'z'

    dustin committed Jun 2, 2010
    * z:
      Achieved still yet deeper understanding of CRLF handling in memcached.
      Reserve CRLF space in items.
      Flush support.
      Rewrote the tests to tell us more about where they fail.
      Fix hash_table_test to not assume it needs to truncate data.
      Stop appending CRLF to items.
    
    Conflicts:
    	ep.cc
    	ep_testsuite.c
    	sqlite-kvstore.cc
    	t/hash_table_test.cc
    
    Change-Id: I99e2f2f9f486c3c64198a25c257b2baff9a73fbc
  2. @dustin

    Achieved still yet deeper understanding of CRLF handling in memcached.

    dustin committed Jun 2, 2010
      memcached asks for two bytes more than it needs and adds its own CRLF.
    
      The engine should store the things it needs only.
    
      *EXCEPT*
    
      incr has to always add CRLF to the end of the value it stores.
    
    The consequence of this is that we will persist two bytes more than
    was strictly required for every record.  The value is opaque to us, so
    we're just going to blindly store what memcached told us because *it*
    deals with the complexity around the additional CRLF (unless it's an
    incr or decr).
    
    Change-Id: I3c6618a596fdd0721589f7a413c5ffc2684a2986
    Reviewed-on: http://review.northscale.com:8080/274
    Reviewed-by: Sean Lynch <seanl@literati.org>
    Tested-by: Eric Lambert <eric.d.lambert@gmail.com>
    Tested-by: Dustin Sallings <dustin@spy.net>
Commits on Jun 2, 2010
  1. @dustin @trondn

    Reserve CRLF space in items.

    dustin committed with trondn Jun 2, 2010
    memcached assumes that the items are all two bytes larger than they
    claim to be, and the extra two bytes are exactly CRLF.
    
    Change-Id: I0f65a1e43399005dbade6b63734990bc69761cf2
    Reviewed-on: http://review.northscale.com:8080/268
    Reviewed-by: Trond Norbye <trond.norbye@gmail.com>
    Tested-by: Trond Norbye <trond.norbye@gmail.com>
Commits on Jun 1, 2010
  1. @dustin

    Stop appending CRLF to items.

    dustin committed Jun 1, 2010
    Change-Id: I9ad5da3fa8547c1691c606f71ae822e78bc1529e
    Reviewed-on: http://review.northscale.com:8080/258
    Tested-by: Trond Norbye <trond.norbye@gmail.com>
    Reviewed-by: Trond Norbye <trond.norbye@gmail.com>
  2. @dustin @trondn

    Stop appending CRLF to items.

    dustin committed with trondn Jun 1, 2010
    Change-Id: I9ad5da3fa8547c1691c606f71ae822e78bc1529e
    Reviewed-on: http://review.northscale.com:8080/258
    Tested-by: Trond Norbye <trond.norbye@gmail.com>
    Reviewed-by: Trond Norbye <trond.norbye@gmail.com>
Commits on May 27, 2010
  1. @dustin @seanlynch

    Using vbuckets (almost) everywhere, but only handling vbucket 0.

    dustin committed with seanlynch May 26, 2010
    Change-Id: I6e2dd55554eabda512ad36560e564015bab016c5
    Reviewed-on: http://review.northscale.com:8080/183
    Reviewed-by: Trond Norbye <trond.norbye@gmail.com>
    Tested-by: Trond Norbye <trond.norbye@gmail.com>
Commits on May 24, 2010
  1. @dustin

    Use atomics for cas counter.

    dustin committed May 24, 2010
    Change-Id: I316f29b55cd43e6be62200f6e1e1f1983780dff2
    Reviewed-on: http://review.northscale.com:8080/149
    Reviewed-by: Trond Norbye <trond.norbye@gmail.com>
    Tested-by: Dustin Sallings <dustin@spy.net>
  2. @dustin @trondn

    Use a LockHolder whenever a lock should be held.

    dustin committed with trondn May 23, 2010
    This prevents issues where a lock is mistakenly released twice because
    LockHolder doesn't know the lock isn't held by the current thread.
    
    Change-Id: I8e701038145e2a64c73a53d24051cb578bc93d0d
    Reviewed-on: http://review.northscale.com:8080/144
    Reviewed-by: Trond Norbye <trond.norbye@gmail.com>
    Tested-by: Trond Norbye <trond.norbye@gmail.com>
Commits on May 15, 2010
  1. @dustin

    Some compile fixes for ubuntu.

    dustin committed May 14, 2010
    Change-Id: I0e2a83262360bcff87afeff5b2c45a0c359ef1a5
    Reviewed-on: http://review.northscale.com:8080/64
    Reviewed-by: Dustin Sallings <dustin@spy.net>
    Tested-by: Dustin Sallings <dustin@spy.net>
Commits on May 11, 2010
  1. @dustin

    Compilation fixes for Linux.

    dustin committed May 10, 2010
    Tested on Ubuntu 10.04 / g++ 4.4.3
    
    Change-Id: Ic7ea591a5391b50d1e2316ccbf3ddbeb0fbefde6
    Reviewed-on: http://review.northscale.com:8080/40
    Reviewed-by: Trond Norbye <trond.norbye@gmail.com>
    Tested-by: Trond Norbye <trond.norbye@gmail.com>
Something went wrong with that request. Please try again.