The most expensive part of packing a repository for transport to
another system is enumerating all of the objects in the repository.
Once this gets to the size of the linux-2.6 repository (1.8 million
objects), enumeration can take several CPU minutes and costs a lot
of temporary working set memory.
Teach PackWriter to efficiently reuse an existing "cached pack"
by answering a clone request with a thin pack followed by a larger
cached pack appended to the end. This requires the repository
owner to first construct the cached pack by hand, and record the
tip commits inside of $GIT_DIR/objects/info/cached-packs:
cd $GIT_DIR
root=$(git rev-parse master)
tmp=objects/.tmp-$$
names=$(echo $root | git pack-objects --keep-true-parents --revs $tmp)
for n in $names; do
chmod a-w $tmp-$n.pack $tmp-$n.idx
touch objects/pack/pack-$n.keep
mv $tmp-$n.pack objects/pack/pack-$n.pack
mv $tmp-$n.idx objects/pack/pack-$n.idx
done
(echo "+ $root";
for n in $names; do echo "P $n"; done;
echo) >>objects/info/cached-packs
git repack -a -d
When a clone request needs to include $root, the corresponding
cached pack will be copied as-is, rather than enumerating all of
the objects that are reachable from $root.
For a linux-2.6 kernel repository that should be about 376 MiB,
the above process creates two packs of 368 MiB and 38 MiB[1].
This is a local disk usage increase of ~26 MiB, due to reduced
delta compression between the large cached pack and the smaller
recent activity pack. The overhead is similar to 1 full copy of
the compressed project sources.
With this cached pack in hand, JGit daemon completes a clone request
in 1m17s less time, but a slightly larger data transfer (+2.39 MiB):
Before:
remote: Counting objects: 1861830, done
remote: Finding sources: 100% (1861830/1861830)
remote: Getting sizes: 100% (88243/88243)
remote: Compressing objects: 100% (88184/88184)
Receiving objects: 100% (1861830/1861830), 376.01 MiB | 19.01 MiB/s, done.
remote: Total 1861830 (delta 4706), reused 1851053 (delta 1553844)
Resolving deltas: 100% (1564621/1564621), done.
real 3m19.005s
After:
remote: Counting objects: 1601, done
remote: Counting objects: 1828460, done
remote: Finding sources: 100% (50475/50475)
remote: Getting sizes: 100% (18843/18843)
remote: Compressing objects: 100% (7585/7585)
remote: Total 1861830 (delta 2407), reused 1856197 (delta 37510)
Receiving objects: 100% (1861830/1861830), 378.40 MiB | 31.31 MiB/s, done.
Resolving deltas: 100% (1559477/1559477), done.
real 2m2.938s
Repository owners can periodically refresh their cached packs by
repacking their repository, folding all newer objects into a larger
cached pack. Since repacking is already considered to be a normal
Git maintenance activity, this isn't a very big burden.
[1] In this test $root was set back about two weeks.
Change-Id: Ib87131d5c4b5e8c5cacb0f4fe16ff4ece554734b
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
CGit pack-objects displays a totals line after the pack data
was fully written. This can be useful to understand some of
the decisions made by the packer, and has been a great tool
for helping to debug some of that code.
Track some of the basic values, and send it to the client when
packing is done:
remote: Counting objects: 1826776, done
remote: Finding sources: 100% (55121/55121)
remote: Getting sizes: 100% (25654/25654)
remote: Compressing objects: 100% (11434/11434)
remote: Total 1861830 (delta 3926), reused 1854705 (delta 38306)
Receiving objects: 100% (1861830/1861830), 386.03 MiB | 30.32 MiB/s, done.
Change-Id: If3b039017a984ed5d5ae80940ce32bda93652df5
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
It isn't strictly necessary to validate every reference's target
object is reachable in the repository before advertising it to a
client. This is an expensive operation when there are thousands of
references, and its very unlikely that a reference uses a missing
object, because garbage collection proceeds from the references and
walks down through the graph. So trying to hide a dangling reference
from clients is relatively pointless.
Even if we are trying to avoid giving a client a corrupt repository,
this simple check isn't sufficient. It is possible for a reference to
point to a valid commit, but that commit to have a missing blob in its
root tree. This can be caused by staging a file into the index,
waiting several weeks, then committing that file while also racing
against a prune. The prune may delete the blob, since its
modification time is more than 2 weeks ago, but retain the commit,
since its modification time is right now.
Such graph corruption is already caught during PackWriter as it
enumerates the graph from the client's want list and digs back
to the roots or common base. Leave the reference validation also
for that same phase, where we know we have to parse the object to
support the enumeration.
Change-Id: Iee70ead0d3ed2d2fcc980417d09d7a69b05f5c2f
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
An ObjectReader implementation may be very slow for a single object,
but yet support bulk queries efficiently by batching multiple small
requests into a single larger request. This easily happens when the
reader is built on top of a database that is stored on another host,
as the network round-trip time starts to dominate the operation cost.
RevWalk, ObjectWalk, UploadPack and PackWriter are the first major
users of this new bulk interface, with the goal being to support an
efficient way to pack a repository for a fetch/clone client when the
source repository is stored in a high-latency storage system.
Processing the want/have lists is now done in bulk, to remove
the high costs associated with common ancestor negotiation.
PackWriter already performs object reuse selection in bulk, but it
now can also do the object size lookup and object counting phases
with higher efficiency. Actual object reuse, deltification, and
final output are still doing sequential lookups, making them a bit
more expensive to perform.
Change-Id: I4c966f84917482598012074c370b9831451404ee
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
When we are creating a pack the higher level application should be able
to override the PackConfig used, allowing it to control the number of
threads used or how much memory is allocated per writer.
Change-Id: I47795987bb0d161d3642082acc2f617d7cb28d8c
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
We only use these variables once, so just put them at the proper
use site and avoid assigning the local variable. The code is a
bit shorter and the intent is a little bit more clear.
Change-Id: I70d120fb149b612ac93055ea39bc053b8d90a5db
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Update a number of calling sites of RevWalk to ensure the walker's
internal ObjectReader is released after the walk is no longer used.
Because the ObjectReader is likely to hold onto a native resource
like an Inflater, we don't want to leak them outside of their
useful scope.
Where possible we also try to share ObjectReaders across several
walk pools, or between a walker and a PackWriter. This permits
the ObjectReader to actually do some caching if it felt inclined
to do so.
Not everything was updated, we'll probably need to come back and
update even more call sites, but these are some of the biggest
offenders. Test cases in particular aren't updated. My plan is to
move most storage-agnostic tests onto some purely in-memory storage
solution that doesn't do compression.
Change-Id: I04087ec79faeea208b19848939898ad7172b6672
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Move PackWriter progress monitors onto the operations
Rather than taking the ProgressMonitor objects in our constructor and
carrying them around as instance fields, take them as arguments to the
actual time consuming operations we need to run.
Change-Id: I2b230d07e277de029b1061c807e67de5428cc1c4
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Ensure ObjectReader used by PackWriter is released
The ObjectReader API demands that we release the reader when we are
done with it. PackWriter contains a reader, which it uses for the
entire packing session. Expose the release of the reader through
a release method on the writer.
This still doesn't address the RevWalk and TreeWalk users, who
don't correctly release their reader. But its a small step in the
right direction.
Change-Id: I5cb0b5c1b432434a799fceb21b86479e09b84a0a
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Similar to what we did with the file code, move the pack writer
into its own package so the related classes and their package
private methods are hidden from the rest of the library.
Change-Id: Ic1b5c7c8c8d266e90c910d8d68dfc8e93586854f
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Instead of peeling things by hand in application level code, defer
the peeling logic into RevWalk's new peel utility method.
Change-Id: Idabd10dc41502e782f6a2eeb56f09566b97775a8
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Under smart HTTP the biDirectionalPipe flag is false, and we return
back immediately at this point in the negotiation process. There is
no need to flush the stream to the client, the request is over and
it will be automatically flushed out by the higher level servlet
that invoked us. Avoiding flush here allows us to only use flush
after a progress message is sent during pack generation.
Change-Id: Id0c8b7e95e3be6ca4c1b479e096bed6b0283b828
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
The strings are externalized into the root resource bundles.
The resource bundles are stored under the new "resources" source
folder to get proper maven build.
Strings from tests are, in general, not externalized. Only in
cases where it was necessary to make the test pass the strings
were externalized. This was typically necessary in cases where
e.getMessage() was used in assert and the exception message was
slightly changed due to reuse of the externalized strings.
Change-Id: Ic0f29c80b9a54fcec8320d8539a3e112852a1f7b
Signed-off-by: Sasa Zivkov <sasa.zivkov@sap.com>
Reduce multi-level buffered streams in transport code
Some transports actually provide stream buffering on their own,
without needing to be wrapped up inside of a BufferedInputStream in
order to smooth out system calls to read or write. A great example
of this is the JSch SSH client, or the Apache MINA SSHD server.
Both use custom buffering to packetize the streams into the encrypted
SSH channel, and wrapping them up inside of a BufferedInputStream
or BufferedOutputStream is relatively pointless.
Our SideBandOutputStream implementation also provides some fairly
large buffering, equal to one complete side-band packet on the main
data channel. Wrapping that inside of a BufferedOutputStream just to
smooth out small writes from PackWriter causes extra data copies, and
provides no advantage. We can save some memory and some CPU cycles
by letting PackWriter dump directly into the SideBandOutputStream's
internal buffer array.
Instead we push the buffering streams down to be as close to the
network socket (or operating system pipe) as possible. This allows
us to smooth out the smaller reads/writes from pkt-line messages
during advertisement and negotation, but avoid copying altogether
when the stream switches to larger writes over a side band channel.
Change-Id: I2f6f16caee64783c77d3dd1b2a41b3cc0c64c159
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Instead of relying on our callers to wrap us up inside of a
BufferedOutputStream and using the proper block sizing, do the
buffering directly inside of SideBandOutputStream. This ensures
we don't get large write-throughs from BufferedOutputStream that
might overflow the configured packet size.
The constructor of SideBandOutputStream is also beefed up to check
its arguments and ensure they are within acceptable ranges for the
current side-band protocol.
Change-Id: Ic14567327d03c9e972f9734b8228178bc448867d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Add a RefFilter interface to ReceivePack and UploadPack
When a user of ReceivePack or UploadPack wants to control what refs
are sent to the client, for instance when some refs should be hidden
from some clients, this interface can be extended to provide a fine
grained control over what refs are sent to the client.
Change-Id: Ie6320b0f8922e1a5e1bad91c016bd476ea094366
Optimize RefAdvertiser performance by avoiding sorting
Don't copy and sort the set of references if they are passed through
in a RefMap or a SortedMap using the key's natural sort ordering.
Either map is already in the order we want to present the items
to the client in, so copying and sorting is a waste of local CPU
and memory.
Change-Id: I49ada7c1220e0fc2a163b9752c2b77525d9c82c1
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Expose RefAdvertiser for reuse outside of the transport package
By making this class and its methods public, and the actual writing
abstract, we can reuse this code for other formats like writing an
info/refs file for HTTP transports.
Change-Id: Id0e349c30a0f5a8c1527e0e7383b80243819d9c5
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Teach UploadPack how to use an RPC style interface
If biDirectionalPipe is false UploadPack does not start out with
the advertisement but instead assumes it should read one block of
want/have lines, process that, and write the ACK/NAKs out.
This means it only is doing one read through the input followed by
one write to the output, which fits with the HTTP request processing
model, and any other type of RPC system.
Change-Id: Ia9f7c46ee556f996367180f15d2caa8572cdd59f
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
The multi_ack_detailed extension breaks out the "ACK %s continue" status
code into "ACK %s common" and "ACK %s ready" states, making it easier to
discover which objects are truely common, and which objects are simply
on a chain the server doesn't care learning about.
Change-Id: Ie8e907424cfbbba84996ca205d49eacf339f9d04
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
Per CQ 3448 this is the initial contribution of the JGit project
to eclipse.org. It is derived from the historical JGit repository
at commit 3a2dd9921c.
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>