PoC of "additional layer store" of Podman/CRI-O and lazy pulling of zstd:chunked #281
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Signed-off-by: Kohei Tokunaga <[email protected]>
AkihiroSuda
reviewed
Mar 5, 2021
| @@ -66,6 +67,11 @@ When '--all-platforms' is given all images in a manifest list must be available. | |||
| Usage: "eStargz chunk size", | |||
| Value: 0, | |||
| }, | |||
| // zstd:chunked flags | |||
| cli.BoolFlag{ | |||
| Name: "zstdchunked", | |||
There was a problem hiding this comment.
nit: zstdchunked -> zstd-chunked (just my preference)
AkihiroSuda
reviewed
Mar 5, 2021
|
|
||
| const ( | ||
| defaultLogLevel = logrus.InfoLevel | ||
| defaultRootDir = "/var/lib/registry-storage" |
There was a problem hiding this comment.
Shouldn't this be something like /var/lib/stargz-registry-storage?
Shouldn't the exec name be like |
ktock
force-pushed
the
als-pool-example
branch
6 times, most recently
from
3582d54
to
e110f40
Compare
Signed-off-by: Kohei Tokunaga <[email protected]>
ktock
added a commit
to ktock/stargz-snapshotter
that referenced
this pull request
Mar 26, 2021
…zstd:chunked) This is the subset of containerd#281. Initially, eStargz is based on gzip compression. But, through zstd:chunked work, it turned out that eStargz is not limited to gzip compression and the same chunking & verifying & prefetching method can be applied to other compression algorithms as well (e.g. zstd). This commit makes `estargz` pkg configurable and agnostic about compression algorithms. For supporting non-gzip compression, the user must implement `estargz.Decompressor` and `estargz.Compressor` interfaces and must plug them to `estargz` tools (e.g. `estargz.Open` and `estargz.NewWriterWithCompression`). `estargz` also provides test suite that is usable for testing these non-gzip eStargz implementations. This commit comes with `zstdchunked` pkg that support zstd compression for eStargz (a.k.a. zstd:chunked), based on the above extensibility. `zstdchunked` pkg contains `zstdchunked.Decompressor` and `zstdchunked.Compressor` that allows `estargz` pkg to use zstd compression (i.e. zstd:chunked) instead of gzip. Layer converter and filesystem now support zstd:chunked leveraging `zstdchunked` pkg. `ctr-remote image optimize` and `ctr-remote image convert` support `--zstdchunked` option that omits zstd-based eStargz and filesystem supports zstd-based eStargz layers by default. Signed-off-by: Kohei Tokunaga <[email protected]>
ktock
added a commit
to ktock/stargz-snapshotter
that referenced
this pull request
Mar 26, 2021
…zstd:chunked) This is the subset of containerd#281. Initially, eStargz is based on gzip compression. But, through zstd:chunked work, it turned out that eStargz is not limited to gzip compression and the same chunking & verifying & prefetching method can be applied to other compression algorithms as well (e.g. zstd). This commit makes `estargz` pkg configurable and agnostic about compression algorithms. For supporting non-gzip compression, the user must implement `estargz.Decompressor` and `estargz.Compressor` interfaces and must plug them to `estargz` tools (e.g. `estargz.Open` and `estargz.NewWriterWithCompression`). `estargz` also provides test suite that is usable for testing these non-gzip eStargz implementations. This commit comes with `zstdchunked` pkg that support zstd compression for eStargz (a.k.a. zstd:chunked), based on the above extensibility. `zstdchunked` pkg contains `zstdchunked.Decompressor` and `zstdchunked.Compressor` that allows `estargz` pkg to use zstd compression (i.e. zstd:chunked) instead of gzip. Layer converter and filesystem now support zstd:chunked leveraging `zstdchunked` pkg. `ctr-remote image optimize` and `ctr-remote image convert` support `--zstdchunked` option that omits zstd-based eStargz and filesystem supports zstd-based eStargz layers by default. Signed-off-by: Kohei Tokunaga <[email protected]>
ktock
added a commit
to ktock/stargz-snapshotter
that referenced
this pull request
Mar 26, 2021
…zstd:chunked) This is the subset of containerd#281. Initially, eStargz is based on gzip compression. But, through zstd:chunked work, it turned out that eStargz is not limited to gzip compression and the same chunking & verifying & prefetching method can be applied to other compression algorithms as well (e.g. zstd). This commit makes `estargz` pkg configurable and agnostic about compression algorithms. For supporting non-gzip compression, the user must implement `estargz.Decompressor` and `estargz.Compressor` interfaces and must plug them to `estargz` tools (e.g. `estargz.Open` and `estargz.NewWriterWithCompression`). `estargz` also provides test suite that is usable for testing these non-gzip eStargz implementations. This commit comes with `zstdchunked` pkg that support zstd compression for eStargz (a.k.a. zstd:chunked), based on the above extensibility. `zstdchunked` pkg contains `zstdchunked.Decompressor` and `zstdchunked.Compressor` that allows `estargz` pkg to use zstd compression (i.e. zstd:chunked) instead of gzip. Layer converter and filesystem now support zstd:chunked leveraging `zstdchunked` pkg. `ctr-remote image optimize` and `ctr-remote image convert` support `--zstdchunked` option that omits zstd-based eStargz and filesystem supports zstd-based eStargz layers by default. Signed-off-by: Kohei Tokunaga <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR contains the following:
Zstd:chunked is a variant of zstd that enables extracting each file separately. The structure based on stargz and its verification logic is based on eStargz. Currently, this is enabled by directly patching
estargzlib for supporting that compression algorithm (zstd) and that footer format. Maybe we should makeestargzlib configurable against compression algorithms and separate zstd:chunked patches into an indipendent package."Additional layer store" is a proposed functionality for https://github.com/containers/storage to provide remotely mounted layers to the runtime. Currently, this is implemented as a thin wrapper of our
fspackage. Instead of implementing snapshotter API, this directly exposes layers on the filesystem. This seems a bit out of scope as a non-core subproject of containerd. So this should be separated to another project, maybe?The following command mounts the store at the
<mountpoint>:This store exposes the following structure required as an additional layer store:
There are also other APIs for debugging:
HelloBench result of Podman + {eStargz, zstd:chunked} on GitHub Actions (
Standard_DS2_v2instance ateastus2):https://github.com/ktock/stargz-snapshotter/actions/runs/623653929
I'll work on experiment on a richer instance with larger number of samples.