This change teaches JGit to use a single commit-graph file to
accelerate commit graph walks. When new commits are added to the
repo, the commit-graph gets further and further behind, and then we
write the whole commit-graph file again to include new commits.
Multiple commit-graph files are not supported in the current
implementation. With this feature, git will write a commit-graph after
every git fetch command that downloads a pack-file from a remote.
JGit can read a commit-graph file from a buffered stream, which means
that we can provide this feature for both FileRepository and
DfsRepository.
Bug: 574368
Change-Id: Ib5c0d6678cb242870a0f5841bd413ad3885e95f6
Signed-off-by: kylezhao <kylezhao@tencent.com>
Update
- com.google.gson to 2.8.7.v20210624-1215
- javaewah to 1.1.12.v20210622-2206
- org.apache.sshd.osgi to 2.7.0.v20210623-0618
- org.apache.sshd.sftp to 2.7.0.v20210623-0618
- org.tukaani.xz to 1.9.0.v20210624-1259
- Apache MINA sshd to 2.7.0
- Remove work-arounds for problems resolved upstream since 2.6.0, and
adapt to upstream API changes.
- update DEPENDENCIES.
CQ: 23469
CQ: 23470
CQ: 23496
CQ: 23497
CQ: 23498
Bug: 574220
Change-Id: I898b216c3492f8488fbf25fa4b49f1250f86f3c8
Also-by: David Ostrovsky <david@ostrovsky.org>
Signed-off-by: Thomas Wolf <thomas.wolf@paranor.ch>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
Remove use of deprecated getAllRefs() in ReceivePack
Repository.getAllRefs() is deprecated and should not
be used anymore. Leverage the ref-db and the retrieval
of refs by prefix and adapt the result to the expected
refname/ref map.
Bug: 534731
Change-Id: I37a9092859f220ddc4e5063d01544f3e82208be8
Signed-off-by: Luca Milanesio <luca.milanesio@gmail.com>
searchForReuse might impact performance in large repositories
The search for reuse phase for *all* the objects scans *all*
the packfiles, looking for the best candidate to serve back to the
client.
This can lead to an expensive operation when the number of
packfiles and objects is high.
Add parameter "pack.searchForReuseTimeout" to limit the time spent
on this search.
Change-Id: I54f5cddb6796fdc93ad9585c2ab4b44854fa6c48
Retry loose object read upon "Stale file handle" exception
When reading loose objects over NFS it is possible that the OS syscall
would fail with ESTALE errors: This happens when the open file
descriptor no longer refers to a valid file.
Notoriously it is possible to hit this scenario when git data is shared
among multiple clients, for example by multiple gerrit instances in HA.
If one of the two clients performs a GC operation that would cause the
packing and then the pruning of loose objects, the other client might
still hold a reference to those objects, which would cause an exception
to bubble up the stack.
The Linux NFS FAQ[1] (at point A.10), suggests that the proper way to
handle such ESTALE scenarios is to:
"[...] close the file or directory where the error occurred, and reopen
it so the NFS client can resolve the pathname again and retrieve the new
file handle."
In case of a stale file handle exception, we now attempt to read the
loose object again (up to 5 times), until we either succeed or encounter
a FileNotFoundException, in which case the search can continue to
Packfiles and alternates.
The limit of 5 provides an arbitrary upper bounds that is consistent to
the one chosen when handling stale file handles for packed-refs
files (see [2] for context).
[1] http://nfs.sourceforge.net/
[2] https://git.eclipse.org/r/c/jgit/jgit/+/54350
Bug: 573791
Change-Id: I9950002f772bbd8afeb9c6108391923be9d0ef51
Fix garbage collection failing to delete pack file
The loosen() method has opened pack file and the open pack file handle
may prevent it from being deleted e.g. on Windows. Fix this by closing
the pack file only after loosen() finished.
Bug: 574178
Change-Id: Icd59931a218d84c9c97b450eea87b21ed01248ff
Signed-off-by: andrew.xian2000@gmail.com
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
Fix PathSuffixFilter: can decide only on full paths
On a subtree, a PathSuffixFilter must return -1 ("indeterminate"),
not 0 ("include"), otherwise negation goes wrong: an indeterminate
result (-1) is passed on, but a decision (0/1) is inverted.
As a result a negated PathSuffixFilter would skip all folders.
Bug: 574253
Change-Id: I27fe785c0d772392a5b5efe0a7b1c9cafcb6e566
Signed-off-by: Thomas Wolf <thomas.wolf@paranor.ch>
Apache MINA sshd doesn't log all values, but it'd be very helpful to
debug KEX problems.
Change-Id: I5a19023c929f39caccde489d51a6fdca711fe5ff
Signed-off-by: Thomas Wolf <thomas.wolf@paranor.ch>
The baseline for the 5.13 release is 5.12.0.202106070339-r.
Change-Id: Ic4730ca296117d7a7f6e86228ce2393a748c3c06
Signed-off-by: Thomas Wolf <thomas.wolf@paranor.ch>
That commit was submitted on master between the 5.12.0 release and
the 5.13.0 version bump.
Change-Id: I679e818bfc5a4695b66548add9a83a22c89a4ffc
Signed-off-by: Thomas Wolf <thomas.wolf@paranor.ch>
Teach independent negotiation (no pack file) using an option "wait-for-done"
From Git commit 9c1e657a8f:
Currently, the packfile negotiation step within a Git fetch cannot be
done independent of sending the packfile, even though there is at
least one application wherein this is useful - push negotiation.
Therefore, make it possible for this negotiation step to be done
independently.
This feature is for protocol v2 only.
In the protocol, the main hindrance towards independent negotiation is
that the server can unilaterally decide to send the packfile. This is
solved by a "wait-for-done" argument: the server will then wait for
the client to say "done". In practice, the client will never say it;
instead it will cease requests once it is satisfied.
Advertising the server capability option "wait-for-done" is behind the
transport config: uploadpack.advertisewaitfordone, which by default is
false.
Change-Id: I5ebd3e99ad76b8943597216e23ced2ed38eb5224
kylezhao
Thomas Wolf
Matthias Sohn
Luca Milanesio
Fabio Ponciroli
Antonio Barone
Marco Miller
andrewxian2000
Ronald Bhuleskar