In the DFS implementation, flushing an inserter writes a new pack to
the storage system and is potentially very slow, but was the only way
to ensure previously-inserted objects were available.  For some tasks,
like performing a series of three-way merges, the total size of all
inserted objects may be small enough to avoid flushing the in-memory
buffered data.

DfsOutputStream already provides a read method to read back from the
not-yet-flushed data, so use this to provide an ObjectReader in the
DFS case.

In the file-backed case, objects are written out loosely on the fly,
so the implementation can just return the existing WindowCursor.

Change-Id: I454fdfb88f4d215e31b7da2b2a069853b197b3dd

Since 2badedcbe0 in-core merges can write up to 10 MiB
into a TemporaryBuffer.Heap strategy, where the data is stored
as a chain of byte[] blocks.

Support the inserter reading up to the streamFileThreshold (default 50
MiB) from the supplied input stream and hash the content to determine
if the merged result blob is already present in the repository.  This
allows the inserter to avoid creating duplicate objects in more cases,
reducing repository pack file churn.

Change-Id: I38967e2a0cff14c0a856cdb46a2c8fedbeb21ed5

Instead of forcing the implementation of the DFS backend to handle
making sure the extension bits are set correctly, have the common
callers in JGit set the extension at the same time they supply the
file sizes to the pack description. This simplifies assumptions for
an implementation of the DFS backend.

Change-Id: I55142ad8ea08a3e2e8349f72b3714578eba9c342

This breaks all existing callers once. Applications are not supposed
to build against the internal storage API unless they can accept API
churn and make necessary updates as versions change.

Change-Id: I2ab1327c202ef2003565e1b0770a583970e432e9

Update the dfs and file GC implementations to prepare and write
bitmaps on the packs that contain the full closure of the object
graph. Update the DfsPackDescription to include the index version.

Change-Id: I3f1421e9cd90fe93e7e2ef2b8179ae2f1ba819ed

PackConstants previously contained string values for the pack and pack
index extension. Change PackConstant to be PackExt, a typed wrapper
around the string pack file extension.

Change-Id: I86ac4db6da8f33aa42d6f37cfcc119e819444318

Previously the size getters and setters had explicit methods for index
and pack. Update the api to be based on the file extension. This will
make it possible to support other extensions in the future, such as
the forthcoming bitmap extensions.

Change-Id: Iab9d4abe0af65b2fc71ad71ef1db0feb6b3b5c58

Previously, the DfsObjDatabase had a hardcoded getPackFile() and
getPackIndex() methods which opens a .pack and .idx file, respectively.
A future change to add a bitmap index will need to be stored in a
parallel .bitmap file. Update the DfsObjDatabase to support opening and
writing of files for any pack extension.

Change-Id: I7c403b501e242096a2d435f6865d6025a9f86108

Change-Id: I3b54cb9f73cb433c71a441a11ddc74cfecdaa1dc

This reverts commit 88fe2836ed.

Auto CRLF isn't special enough to be screwing around with the buffers
used for raw byte processing of the ObjectInserter API. If it needs a
buffer to process a file that is bigger than the buffer allocated by
an ObjectInserter, it needs to do its own buffer management.

Change-Id: Ida4aaa80d0f9f78035f3d2a9ebdde904c980f89a

CRLF only works for small files, where small is the size of the
buffer, i.e. about 8K. This QD fix reallocates the buffer to be
large enough.

Bug: 369780
Change-Id: Ifc34ad204fbf5986b257a5c616e4a8c601e8261a

In practice the DHT storage layer has not been performing as well as
large scale server environments want to see from a Git server.

The performance of the DHT schema degrades rapidly as small changes
are pushed into the repository due to the chunk size being less than
1/3 of the pushed pack size.  Small chunks cause poor prefetch
performance during reading, and require significantly longer prefetch
lists inside of the chunk meta field to work around the small size.

The DHT code is very complex (>17,000 lines of code) and is very
sensitive to the underlying database round-trip time, as well as the
way objects were written into the pack stream that was chunked and
stored on the database.  A poor pack layout (from any version of C Git
prior to Junio reworking it) can cause the DHT code to be unable to
enumerate the objects of the linux-2.6 repository in a completable
time scale.

Performing a clone from a DHT stored repository of 2 million objects
takes 2 million row lookups in the DHT to locate the OBJECT_INDEX row
for each object being cloned. This is very difficult for some DHTs to
scale, even at 5000 rows/second the lookup stage alone takes 6 minutes
(on local filesystem, this is almost too fast to bother measuring).
Some servers like Apache Cassandra just fall over and cannot complete
the 2 million lookups in rapid fire.

On a ~400 MiB repository, the DHT schema has an extra 25 MiB of
redundant data that gets downloaded to the JGit process, and that is
before you consider the cost of the OBJECT_INDEX table also being
fully loaded, which is at least 223 MiB of data for the linux kernel
repository.  In the DHT schema answering a `git clone` of the ~400 MiB
linux kernel needs to load 248 MiB of "index" data from the DHT, in
addition to the ~400 MiB of pack data that gets sent to the client.
This is 193 MiB more data to be accessed than the native filesystem
format, but it needs to come over a much smaller pipe (local Ethernet
typically) than the local SATA disk drive.

I also never got around to writing the "repack" support for the DHT
schema, as it turns out to be fairly complex to safely repack data in
the repository while also trying to minimize the amount of changes
made to the database, due to very common limitations on database
mutation rates..

This new DFS storage layer fixes a lot of those issues by taking the
simple approach for storing relatively standard Git pack and index
files on an abstract filesystem. Packs are accessed by an in-process
buffer cache, similar to the WindowCache used by the local filesystem
storage layer. Unlike the local file IO, there are some assumptions
that the storage system has relatively high latency and no concept of
"file handles". Instead it looks at the file more like HTTP byte range
requests, where a read channel is a simply a thunk to trigger a read
request over the network.

The DFS code in this change is still abstract, it does not store on
any particular filesystem, but is fairly well suited to the Amazon S3
or Apache Hadoop HDFS. Storing packs directly on HDFS rather than
HBase removes a layer of abstraction, as most HBase row reads turn
into an HDFS read.

Most of the DFS code in this change was blatently copied from the
local filesystem code. Most parts should be refactored to be shared
between the two storage systems, but right now I am hesistent to do
this due to how well tuned the local filesystem code currently is.

Change-Id: Iec524abdf172e9ec5485d6c88ca6512cd8a6eafb