Commit graph

10 commits

Author SHA1 Message Date
Florian Klink
616fa4476f refactor(tvix/store): remove ChunkService
Whether chunking is involved or not, is an implementation detail of each
Blobstore. Consumers of a whole blob shouldn't need to worry about that.
It currently is not visible in the gRPC interface either. It
shouldn't bleed into everything.

Let the BlobService trait provide `open_read` and `open_write` methods,
which return handles providing io::Read or io::Write, and leave the
details up to the implementation.

This means, our custom BlobReader module can go away, and all the
chunking bits in there, too.

In the future, we might still want to add more chunking-aware syncing,
but as a syncing strategy some stores can expose, not as a fundamental
protocol component.

This currently needs "SyncReadIntoAsyncRead", taken and vendored in from
https://github.com/tokio-rs/tokio/pull/5669.
It provides a AsyncRead for a sync Read, which is necessary to connect
our (sync) BlobReader interface to a GRPC server implementation.

As an alternative, we could also make the BlobReader itself async, and
let consumers of the trait (EvalIO) deal with the async-ness, but this
is less of a change for now.

In terms of vendoring, I initially tried to move our tokio crate to
these commits, but ended up in version incompatibilities, so let's
vendor it in for now.

Change-Id: I5969ebbc4c0e1ceece47981be3b9e7cfb3f59ad0
Reviewed-on: https://cl.tvl.fyi/c/depot/+/8551
Tested-by: BuildkiteCI
Reviewed-by: tazjin <tazjin@tvl.su>
2023-05-11 14:27:57 +00:00
Florian Klink
8263c024c9 refactor(tvix/store/import): use DirectoryPutter in import.rs
This should allow import_path to communicate to a gRPC remote store,
that actually verifies the Directory nodes are interconnected.

Change-Id: Ic5d28c33518f50dedec15f1732d81579a3afaff1
Reviewed-on: https://cl.tvl.fyi/c/depot/+/8357
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: tazjin <tazjin@tvl.su>
Tested-by: BuildkiteCI
2023-04-07 09:26:36 +00:00
Florian Klink
ee23220564 refactor(tvix/store/directorysvc): use [u8; 32] instead of Vec<u8>
Also, simplify the trait interface, only allowing lookups of Directory
objects by their digest.

Change-Id: I6eec28a8cb0557bed9b69df8b8ff99a5e0f8fe35
Reviewed-on: https://cl.tvl.fyi/c/depot/+/8313
Tested-by: BuildkiteCI
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: tazjin <tazjin@tvl.su>
2023-03-16 13:47:32 +00:00
Florian Klink
7ffb2676ee refactor(tvix/store): add read_all_and_chunk method
This moves the logic from src/import.rs that

 - reads over the contents of a file
 - chunks them up and uploads individual chunks
 - keeps track of the uploaded chunks in a BlobMeta structure
 - returns the hash of the blob and the BlobMeta structure

… into a generic read_all_and_chunk function in
src/chunkservice/util.rs.

It will work on anything implementing io::Read, not just files, which
will help us in a bit.

Change-Id: I53bf628114b73ee2e515bdae29974571ea2b6f6f
Reviewed-on: https://cl.tvl.fyi/c/depot/+/8259
Reviewed-by: tazjin <tazjin@tvl.su>
Tested-by: BuildkiteCI
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Autosubmit: flokli <flokli@flokli.de>
2023-03-13 08:46:51 +00:00
Florian Klink
e4687765d7 refactor(tvix/store): bump fastcdc dep
This removes the use of Box::new, switching fastcdc to version 3.0.2
with https://github.com/nlfiedler/fastcdc-rs/issues/25 fixed.

Change-Id: I64f388b9e0a7f358e25a8bb7ca0e4df1d3bb01c4
Reviewed-on: https://cl.tvl.fyi/c/depot/+/8249
Tested-by: BuildkiteCI
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Reviewed-by: tazjin <tazjin@tvl.su>
Autosubmit: flokli <flokli@flokli.de>
2023-03-11 14:12:46 +00:00
Florian Klink
b049b88d2d refactor(tvix/store): factor out hash update into function
We're using this in a bunch of places. Let's move it into a helper
function.

Change-Id: I118fba35f6d343704520ba37280e4ca52a61da44
Reviewed-on: https://cl.tvl.fyi/c/depot/+/8251
Autosubmit: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
2023-03-11 14:12:45 +00:00
Florian Klink
ead113cdfc feat(tvix/store/import): use StreamCDC instead of blobwriter
This seems to be way faster.

Change-Id: Ica7cee95d108c51fe67365f07366634ddbbfa060
Reviewed-on: https://cl.tvl.fyi/c/depot/+/8246
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Reviewed-by: tazjin <tazjin@tvl.su>
Autosubmit: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
2023-03-11 12:54:07 +00:00
Florian Klink
510927e43a feat(tvix/store): use rayon to upload chunks concurrently
Look at the data that's written to us, and upload all chunks but the
rest in parallel, using rayon. This required moving `upload_chunk`
outside the struct, and accepting a ChunkService to use for upload
(which it was previously getting from `self.chunk_service`).

This doesn't speed up things too much for now, because things are still
mostly linear.

Change-Id: Id785b5705c3392214d2da1a5b6a182bcf5048c8d
Reviewed-on: https://cl.tvl.fyi/c/depot/+/8195
Autosubmit: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
2023-03-10 11:53:29 +00:00
Florian Klink
2ef60282b6 feat(tvix/store/import): make sure entries are sorted
The Directory service does already reject inserting invalid (wrongly
sorted) Directory messages, but our test case didn't provoke it.

Change-Id: I228e201925e8999186659a2d8da0118db184d9ab
Reviewed-on: https://cl.tvl.fyi/c/depot/+/8167
Tested-by: BuildkiteCI
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
2023-03-10 10:58:55 +00:00
Florian Klink
b29d1ae372 feat(tvix/store): add import::import_path
This imports the contents at a given Path into the tvix store.

It doesn't register the contents at a Path in the store itself, that's up
to the PathInfoService.

Change-Id: I2c493532d65b90f199ddb7dfc90249f5c2957dee
Reviewed-on: https://cl.tvl.fyi/c/depot/+/8159
Reviewed-by: raitobezarius <tvl@lahfa.xyz>
Tested-by: BuildkiteCI
2023-03-10 10:58:55 +00:00