Summary: Only four tests remaining after this.
Reviewed By: kulshrax
Differential Revision: D29229656
fbshipit-source-id: 56c0a17f6585263e983ce8bc3c345b1f266422e0
Summary:
Remove packfile-specific parts of tests and modify them to test without depending on packfiles where possible.
Currently debugindexedlogdatastore and debugindexedloghistorystore appear to be broken, and debugdumpindexedlog just dumps the raw indexedlog contents, without any semantic information, so for the time being I've simply removed most packfile inspection.
Reviewed By: DurhamG
Differential Revision: D29099241
fbshipit-source-id: 86c4f9c83520374560587b8bec5c569d9c5c6510
Summary:
We've rolled both of these out to 100%. Let's make this the default so
we can delete those configs.
Reviewed By: quark-zju
Differential Revision: D26233645
fbshipit-source-id: cd7a08c404483f78ab714763870f5bf0fa801e7a
Summary:
For repositories that have the old-style LFS extension enabled, the pointers
are stored in packfiles/indexedlog alongside with a flag that signify to the
upper layers that the blob is externally stored. With the new way of doing LFS,
pointers are stored separately.
When both are enabled, we are observing some interesting behavior where
different get and get_meta calls may return different blobs/metadata for the
same filenode. This may happen if a filenode is stored in both a packfile as an
LFS pointers, and in the LFS store. Guaranteeing that the revisionstore code is
deterministic in this situation is unfortunately way too costly (a get_meta
call would for instance have to fully validate the sha256 of the blob, and this
wouldn't guarantee that it wouldn't become corrupted on disk before calling
get).
The solution take here is to simply ignore all the lfs pointers from
packfiles/indexedlog when remotefilelog.lfs is enabled. This way, there is no
risk of reading the metadata from the packfiles, and the blob from the
LFSStore. This brings however another complication for the user created blobs:
these are stored in packfiles and would thus become unreadable, the solution is
to simply perform a one-time full repack of the local store to make sure that
all the pointers are moved from the packfiles to to LFSStore.
In the code, the Python bindings are using ExtStoredPolicy::Ignore directly as
these are only used in the treemanifest code where no LFS pointers should be
present, the repack code uses ExtStoredPolicy::Use to be able to read the
pointers, it wouldn't be able to otherwise.
Reviewed By: DurhamG
Differential Revision: D22951598
fbshipit-source-id: 0e929708ba5a3bb2a02c0891fd62dae1ccf18204
Summary:
If the LFS server is down, we are going to retry fetching filenode from the
Mercurial server directly, who is expected to not return a pointer.
Previously, this was achieved by adding a hack into `get_missing`, but since
the function is no longer called during prefetch calls, we cannot rely on it
anymore. Instead, we can wrap the regular remote store and translate all the
StoreKey::Content onto their corresponding hgid keys.
Reviewed By: DurhamG
Differential Revision: D22565604
fbshipit-source-id: 2532a1fc3dfd9ba5600957ed5cf905255cb5b3fd
Summary:
In some rare situation, it is possible to have an LFS pointer in both the
packfile and the LFS store. In this case, the flags for this filenode may or
may not be empty, making the assert too strong.
Reviewed By: DurhamG
Differential Revision: D22565605
fbshipit-source-id: b82282b3f47af2a9e607f09a7a7d271ecc4e521a