Use a proper graph library to ensure all nodes are reachable from the
root.
We had a bit of that handrolled during add(), as well as later, which
had an annoying bug:
Redundant nodes were omitted during insert, but when returning the list
during finalize, we did not properly account they need to be introduced
before their parents are sent.
We now simply populate a petgraph DiGraph during insert (skipping
inserting nodes we already saw), and use petgraph's DfsPostOrder to
traverse the graph during finalize.
If the number of returned indices equals the total number of nodes in
the graph, all nodes are reachable from the root, we can consume the
graph and return the nodes as a vec, in the same order as the traversal
(and insertion).
Providing a regression test for the initial bug is challenging, as the
current code uses a bunch of HashSets. I manually tested ingesting a
full NixOS closure using this mechanism (via gRPC, which exposes this
problem, as it validates twice), and it now works.
Change-Id: Ic1d5e3e981f2993cc08c5c6b60ad895e578326dc
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11418
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: Connor Brewster <cbrewster@hey.com>
Tested-by: BuildkiteCI
Have a Option<Box<dyn DirectoryPutter>>, which is lazily initialized
whenever we first want to upload a directory.
Have the loop explicitly break when it encounters the root_node, and
deal with the flushing after the loop.
Deal with the FUTUREWORK (assertion for root directory digest matching
what the DirectoryPutter returns).
Change-Id: Iefc4904d8b8387e868fb752d40e3e4e4218c7407
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11417
Tested-by: BuildkiteCI
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: Connor Brewster <cbrewster@hey.com>
Separate this a bit stronger from the main application flow.
Change-Id: I2e9bd3ec47cc6e37256ba6afc6e0586ddc9a051f
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11416
Autosubmit: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
Reviewed-by: Brian Olsen <me@griff.name>
Reviewed-by: Connor Brewster <cbrewster@hey.com>
Align these names and comments with the two users, to make it more
obvious we're doing the same thing here, just use a different method to
come up with entries_per_depths.
Change-Id: I42058e397588b6b57a6299e87183bef27588b228
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11415
Tested-by: BuildkiteCI
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: Connor Brewster <cbrewster@hey.com>
This did very little, and especially the part of relying on the outside
caller to pass in a Directory if the type is a directory required having
per-entry-type specific logic anyways.
It's cleaner to just inline it.
Change-Id: I997a8513ee91c67b0a2443cb5cd9e8700f69211e
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11414
Autosubmit: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
Reviewed-by: Connor Brewster <cbrewster@hey.com>
This makes it easier to understand the code.
Change-Id: I0a9047433000551a6ba1f50a8c5c93527bc86216
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11413
Tested-by: BuildkiteCI
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: Connor Brewster <cbrewster@hey.com>
We previously updated this.pos also in case the underlying read returned
an error.
Also, use the ready! macro to remove the match block, and instrument
errors returned during start_seek.
Change-Id: Ic32e26579d964a76b45687134acc48d72d67c36f
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11421
Reviewed-by: Brian Olsen <me@griff.name>
Reviewed-by: Connor Brewster <cbrewster@hey.com>
Tested-by: BuildkiteCI
Autosubmit: flokli <flokli@flokli.de>
When a directory or file is open()'ed, we already put some data into
a lookup table, and subsequent operations then use the returned handle
id.
By also adding the span that's been created during these calls into the
lookup table, we can properly set the span parent for these requests,
nicely connecting the individual operations to the bigger picture.
Change-Id: Ia354842fccdbc7f45c2d3efda3acf058b2dbc48e
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11429
Reviewed-by: Connor Brewster <cbrewster@hey.com>
Tested-by: BuildkiteCI
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: Brian Olsen <me@griff.name>
Instead of creating another child span, we can use
`tracing::Span::current().record(k,v)` to add an additional field to the
current span.
Change-Id: I337faac0e73a0da6eb0a52cb75c2e8c026eff774
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11428
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Tested-by: BuildkiteCI
This currently shares some code with readdir, except it's also providing
a second `fuse_backend_rs::api::filesystem::Entry` argument to the
`add_entry` function call.
Refactoring this to reduce some duplication is left for a future CL.
Change-Id: I282c8dfc6a711d00a4482c87cbb84d4950c0aee9
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11426
Tested-by: BuildkiteCI
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Use rq.handle in `release` too, and remove interpolating it into the
log message itself.
Also update the comment, we don't get ownership, just simply drop, and
change the level to warn!, as suggested in cl/11425.
Change-Id: If4e6cff6d8b580671b1548ae3862851db4af6694
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11427
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Tested-by: BuildkiteCI
Autosubmit: flokli <flokli@flokli.de>
Similar to how we already handle opening files, implement opendir/
releasedir, and keep a map of dir_handles. They point to the rx side of
a channel.
This greatly improves performance listing the root of the filesystem
when used inside tvix-store, as we don't need to re-request the listing
(and skip to the desired position) all the time.
Change-Id: I0d3ec4cb70a8792c5a1343439cf47d78d9cbb1d6
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11425
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Tested-by: BuildkiteCI
This allows us acquiring the lock in sync code still. Also, simplify
some of the error handling a bit.
Change-Id: I29e83b715f92808e95ecb0ae9de787339d1a371d
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11424
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Autosubmit: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
We can just pass an async move closure to `self.tokio_handle.block_on`
and make this a bit shorter.
Change-Id: Iba674f34f22ba7a7de7c5bae59d64584884cb17c
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11423
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Tested-by: BuildkiteCI
This exposes `user.tvix.castore.{blob,directory}.digest` xattr keys for
files and directories:
```
❯ getfattr -d /tmp/tvix/06jrrv6wwp0nc1m7fr5bgdw012rfzfx2-nano-7.2-info
getfattr: Removing leading '/' from absolute path names
user.tvix.castore.directory.digest="b3:SuYDcUM9RpWcnA40tYB1BtYpR0xw72v3ymhKDQbBfe4="
❯ getfattr -d /tmp/tvix/156a89x10c3kaby9rgf3fi4k0p6r9wl1-etc-shells
getfattr: Removing leading '/' from absolute path names
user.tvix.castore.blob.digest="b3:pZkwZoHN+/VQ8wkaX0wYVXZ0tV/HhtKlSqiaWDK7uRs="
```
It's currently mostly used for debugging, though it might be useful for
tvix-castore-aware syncing programs using the filesystem too.
Change-Id: I26ac3cb9fe51ffbf7f880519f26741549cb5ab6a
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11422
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Tested-by: BuildkiteCI
Reviewed-by: Brian Olsen <me@griff.name>
We don't need to copy if we explicitly say that the returned
Option<Path> may hold onto bytes from the passed in &DirEntry.
Change-Id: Ib46b6fd2f8f19a45f8bef79c4c1d2fa6b490cad7
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11410
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Tested-by: BuildkiteCI
It was returning `{ line, col, file }` instead of `{ line, column, file
}`.
Note this is not a regression in Nix, Nix always had `column` since
2013.
Change-Id: I4871a201b32324dff0432526d0449806a535828f
Signed-off-by: Ryan Lahfa <tvl@lahfa.xyz>
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11409
Reviewed-by: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
This tells josh to include the patch from
third_party/overlays/patches/cbtemulator-uds.patch at
nixpkgs/cbtemulator-uds.patch.
We then use the `additionalOverlays` feature introduced earlier, and
pass it an overlay adding the patch to `cbtemulator`.
This only gets applied if tvix/shell.nix is not called with a pkgs
argument, which is the case in the tvix view.
Change-Id: I7bc5543460ddd30325ab0998e891c15fcaf8bcfe
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11408
Tested-by: BuildkiteCI
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Autosubmit: flokli <flokli@flokli.de>
While the compiler might realize it's a good idea to inline these,
especially considering this is in the same crate [1], it doesn't hurt to
add that annotation too.
Suggested in cl/11385.
[1]: https://nnethercote.github.io/perf-book/inlining.html
Change-Id: If679fce0e1aab9ec681ba60f689d386d6fd92d61
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11403
Tested-by: BuildkiteCI
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Autosubmit: flokli <flokli@flokli.de>
We bumped nixpkgs past that bump, so we only need to carry our
downstream UDS support patch.
Change-Id: I56504d71a8327cba2e43aa36b7b657f93cf35e51
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11405
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Tested-by: BuildkiteCI
We should cap the time of a `cargo build && cargo test` to a reasonable
amount of time.
Change-Id: If618198459ac267f9f6d7eb60466534a95a95c2c
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11404
Autosubmit: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
This is a stream of DirEntry, so let's call it direntry_stream.
Change-Id: I5b3cb4efba899d746393f75f6ece7eaa79424717
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11401
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Autosubmit: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
Instead of using an additional Nix build producing a directory structure
containing xz-compressed NAR files and NARInfos, and uploading these via
`curl` to NAR-bridge, which then decomposes them and uploads them to
tvix-store, have tvix-store interpret the exported reference graph
directly, and deal with the uploading on its own.
Change-Id: I44ee0b132944c53d11abb688d2d017f0cc9c3d97
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11398
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Autosubmit: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
This allows copying individual store paths from the local /nix/store
into tvix-store.
As we don't support getting this information from Nix yet, we currently
expect metadata to be provided externally:
Nix' `exportReferencesGraph` feature contains pretty much all data we
need for this. Expect a list of this information at a well-known key
(`closure`, similar to nixpkgs' `pkgs/build-support/binary-cache/
default.nix`).
We currently simply upload all store paths sequentially, without any
parallelism or awareness in how the reference graph looks like.
As long as the connected stores don't enforce this, this is fine, at
least for now.
Change-Id: Ib83c998465adddfdb110db994843c44e26b3d3d8
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11397
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Tested-by: BuildkiteCI
Autosubmit: flokli <flokli@flokli.de>
`toJSON` transform a Nix structure into a JSON string.
For each context in that Nix structure, the JSON string must possess it.
Thus, it is necessary to take the union of all contexts and attach it to
the final structure.
Unfortunately, the return type of `into_json` is a serde's JSON object,
not a string. Therefore, it is not possible to reuse `NixString`
machinery.
Context tests are reinforced as Nix does not test those behaviors.
Fixes b/393.
Change-Id: I5afdbc4e18dd70469192c1aa657d1049ba330149
Signed-off-by: Ryan Lahfa <tvl@lahfa.xyz>
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11266
Reviewed-by: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
There's little reason to keep the nar writer using Async{Read,Write}
traits from futures, while everything else async in tvix (and
nix-compat) uses tokio.
Change-Id: I8cd1efcd0dd5bb76471de997603c7b701a5095de
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11391
Tested-by: BuildkiteCI
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Reviewed-by: Brian Olsen <me@griff.name>
This doesn't have much to do with the plain "wire" format, it's merely
one user of it.
Also, use the more "public" `wire::` API to read/write bytes, strings,
bools and u64s.
Change-Id: I98dddcc3004dfde7a0c009958fe84a840f77b188
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11390
Tested-by: BuildkiteCI
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Reviewed-by: Brian Olsen <me@griff.name>
Move everything bytes-related into its own module, and re-export
both bytes and primitive in a flat space from wire/mod.rs.
Expose this if a `wire` feature flag is set. We only have `async` stuff
in here.
Change-Id: Ia4ce4791f13a5759901cc9d6ce6bd6bbcca587c7
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11389
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Tested-by: BuildkiteCI
Reviewed-by: Brian Olsen <me@griff.name>
We don't have it in all Cargo.toml files either.
Change-Id: Ifcb333a39603a7e402ab0068e0f0f1b5af769e13
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11388
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Tested-by: BuildkiteCI
Reviewed-by: Brian Olsen <me@griff.name>
Autosubmit: flokli <flokli@flokli.de>
This is another representation about information Nix has on a Store
Path. It's encountered in the exportReferencesGraph feature.
Change-Id: Ia0c08c369f4af9e50ebe3a5053bd98fadaa0003d
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11396
Tested-by: BuildkiteCI
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
This actually does a bit of allocation. There's two Vecs, one for
references and one for signatures.
We can get rid of the Vec at nar_hash.
Change-Id: Ie025309b6678f83f5b961d49ff75dcfc7da145a1
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11395
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Tested-by: BuildkiteCI
We use the (slightly more tolerant) from_str to deserialize, and
serialize out as SRI.
Change-Id: If76b0ed2d4e243904f02df34f6c90b976c0bab8c
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11393
Tested-by: BuildkiteCI
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Move the code implementing it from StorePath to StorePathRef, and have
the StorePath impls use that too.
Drop the debug_assert in every comparison - we have tests for this to
ensure it keeps working, and built up some confidence by piping a lot of
other store paths through it in the meantime.
Change-Id: I288bad3dfa597f68d63c4bcda7791f722b7a8ced
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11392
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Tested-by: BuildkiteCI
I updated https://github.com/googleapis/google-cloud-go/pull/9665 in the
meantime, and GH decided to GC the patches.
Vendor the patch in for now (manually stripping the `bigtable/` prefix
in the path). Hopefully the PR itself gets merged soon.
Change-Id: I5b7ba78ccaf5c792c1445818b23b52d6f17155a2
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11402
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Tested-by: BuildkiteCI
Autosubmit: flokli <flokli@flokli.de>
Put behind the "cloud" backend, like in the `tvix-castore` crate.
Change-Id: Ib38d198baf11ab2a4b6dc405121676147c424611
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11362
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: Connor Brewster <cbrewster@hey.com>
Tested-by: BuildkiteCI
We forgot to only write the part of the padding that has not been
written already.
Unfortunately it seems the tokio mocking thing does buffer small writes,
so the tests succeeds even with the bug :-/
Change-Id: I5a6cf04212d559728639427c57207094d507ec75
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11384
Tested-by: BuildkiteCI
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Reviewed-by: Brian Olsen <me@griff.name>
Autosubmit: flokli <flokli@flokli.de>
Right now `builtins.hashFile` always reads the entire file into memory
before hashing, which is not ideal for large files. This replaces
`read_to_string` with `open_file` which allows calculating the hash of
the file without buffering it entirely into memory. Other callers can
continue to buffer into memory if they choose, but they still use the
`open_file` VM request and then call `read_to_string` or `read_to_end`
on the `std::io::Reader`.
Fixes b/380
Change-Id: Ifa1c8324bcee8f751604b0b449feab875c632fda
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11236
Reviewed-by: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
This adds a Directory service using
https://cloud.google.com/bigtable/docs/ as a K/V store.
Directory (closures) are put in individual keys.
We don't do any bucketed upload of directory closures (yet), as castore/
fs does query individually, does not request recursively (and buffers).
This will be addressed by store composition at some point.
Change-Id: I7fada45bf386a78b7ec93be38c5f03879a2a6e22
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11212
Tested-by: BuildkiteCI
Reviewed-by: Connor Brewster <cbrewster@hey.com>
Autosubmit: flokli <flokli@flokli.de>