[sui-tool] dump-packages uses GraphQL #18337
Merged
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
|
The latest updates on your projects. Learn more about Vercel for Git ↗︎
3 Skipped Deployments
|
amnn
commented
Jun 20, 2024
|
|
||
| #[derive(cynic::QueryFragment, Debug)] | ||
| pub(crate) struct ServiceConfig { | ||
| pub(crate) max_page_size: i32, |
There was a problem hiding this comment.
cynic (the GraphQL client library that I'm using here) maps Int to i32 -- this is technically correct, according to the spec. We use it to represent values that are less than 2^53 because that's what we can generally rely on being transferred faithfully by JSON across languages, and async-graphql lets us do this as well (maps u64 to Int).
This makes me think that we need to introduce a new scalar, like UInt64 (...or UInt53, or UInt?) that maps to a JSON numeric literal but that we can give this meaning. cc @hayes-mysten as our resident expert -- this is not something that we need for JS' sake, but other strongly typed languages are likely to face this issue, and the overhead of this extra scalar seems minimal to accommodate that.
wlmyng
approved these changes
Jun 21, 2024
| } | ||
|
|
||
| /// Query types related to fetching packages. | ||
| pub(crate) mod packages { |
There was a problem hiding this comment.
did you hand-roll these or did cynic generate these for you?
There was a problem hiding this comment.
I hand-rolled these, although cynic does have a builder UI: https://generator.cynic-rs.dev/
Grouping the types into their own modules, and adding a build function are my own additions, but otherwise I think you'll get something quite similar from the builder.
| /// existing empty directory (in which case it will be filled), or an existing directory that has | ||
| /// been written to in the past (in which case this invocation will pick back up from where the | ||
| /// previous invocation left off). | ||
| pub async fn dump( |
There was a problem hiding this comment.
For my understanding, in other words, dump all the packages that have been created by before_checkpoint
There was a problem hiding this comment.
That's right -- the reason why it's phrased more "passively" (that we're ensuring they exist) is that we use the last-checkpoint file to avoid refetching packages (e.g. if you ran the tool once and then you run it again in the same directory, it will pick up where it left off).
amnn
added a commit
that referenced
this pull request
Jul 7, 2024
## Description Whilst working on #18337, I noticed that we were over-using the `Int` scalar -- using it to represent values that could exceed 2^31 - 1 -- when the GraphQL spec states that `Int` must be a 32-bit signed integer. We made this decision at the time (a) because `async-graphql` allowed converting `u64`s to `Int` and we were primarily concerned with the fact that although JSON doesn't specify a precision for its numeric types, JS (among other languages), assumes it is an IEEE double-precision floating point number, so can only represent integral values precisely below 2^53. `cynic` (a Rust GraphQL client library) is (correctly) stricter, however, and maps an `Int` to an `i32`, always. There may be other similarly strict client libraries for other languages. This PR introduces a new scalar, `UInt`, that maps to a JSON number literal, just like `Int`, but allows us to ascribe our own meaning (in this case, it will be an unsigned number, and it can be as large as 2^53). This scalar has been used in many cases where we had previously used `Int`: sequence numbers, counts of objects, checkpoints, transactions, etc. While other uses continue to use `Int` (pagination limits, service limits, values bounded by the number of validators). In some cases, we have switched from `BigInt` to using this scalar notably: - the db cost estimate, which was previously a `BigInt` because we were unsure of its scale, but in hindsight, after benchmarking, it is unlikely that we would want to set a limit greater than 2^31 - 1. - the number of checkpoints in an epoch, as the number of transactions in an epoch (a number that is guaranteed to be greater) is being represented using an `Int` at the moment (and soon a `UInt`). This will be a breaking change, so will only go out with the new major version. Hopefully, this change will be minimal as the format of this scalar over the wire is the same as for `Int`, but it will require existing clients to register a new scalar in most cases. ## Test plan Existing tests: ``` sui-graphql-rpc$ cargo nextest run sui-graphql-e2e-tests$ cargo nextest run --features pg_integration ```
7 tasks
amnn
added a commit
that referenced
this pull request
Jul 13, 2024
## Description Whilst working on #18337, I noticed that we were over-using the `Int` scalar -- using it to represent values that could exceed 2^31 - 1 -- when the GraphQL spec states that `Int` must be a 32-bit signed integer. We made this decision at the time (a) because `async-graphql` allowed converting `u64`s to `Int` and we were primarily concerned with the fact that although JSON doesn't specify a precision for its numeric types, JS (among other languages), assumes it is an IEEE double-precision floating point number, so can only represent integral values precisely below 2^53. `cynic` (a Rust GraphQL client library) is (correctly) stricter, however, and maps an `Int` to an `i32`, always. There may be other similarly strict client libraries for other languages. This PR introduces a new scalar, `UInt`, that maps to a JSON number literal, just like `Int`, but allows us to ascribe our own meaning (in this case, it will be an unsigned number, and it can be as large as 2^53). This scalar has been used in many cases where we had previously used `Int`: sequence numbers, counts of objects, checkpoints, transactions, etc. While other uses continue to use `Int` (pagination limits, service limits, values bounded by the number of validators). In some cases, we have switched from `BigInt` to using this scalar notably: - the db cost estimate, which was previously a `BigInt` because we were unsure of its scale, but in hindsight, after benchmarking, it is unlikely that we would want to set a limit greater than 2^31 - 1. - the number of checkpoints in an epoch, as the number of transactions in an epoch (a number that is guaranteed to be greater) is being represented using an `Int` at the moment (and soon a `UInt`). This will be a breaking change, so will only go out with the new major version. Hopefully, this change will be minimal as the format of this scalar over the wire is the same as for `Int`, but it will require existing clients to register a new scalar in most cases. ## Test plan Existing tests: ``` sui-graphql-rpc$ cargo nextest run sui-graphql-e2e-tests$ cargo nextest run --features pg_integration ```
amnn
added a commit
that referenced
this pull request
Jul 13, 2024
## Description Whilst working on #18337, I noticed that we were over-using the `Int` scalar -- using it to represent values that could exceed 2^31 - 1 -- when the GraphQL spec states that `Int` must be a 32-bit signed integer. We made this decision at the time (a) because `async-graphql` allowed converting `u64`s to `Int` and we were primarily concerned with the fact that although JSON doesn't specify a precision for its numeric types, JS (among other languages), assumes it is an IEEE double-precision floating point number, so can only represent integral values precisely below 2^53. `cynic` (a Rust GraphQL client library) is (correctly) stricter, however, and maps an `Int` to an `i32`, always. There may be other similarly strict client libraries for other languages. This PR introduces a new scalar, `UInt`, that maps to a JSON number literal, just like `Int`, but allows us to ascribe our own meaning (in this case, it will be an unsigned number, and it can be as large as 2^53). This scalar has been used in many cases where we had previously used `Int`: sequence numbers, counts of objects, checkpoints, transactions, etc. While other uses continue to use `Int` (pagination limits, service limits, values bounded by the number of validators). In some cases, we have switched from `BigInt` to using this scalar notably: - the db cost estimate, which was previously a `BigInt` because we were unsure of its scale, but in hindsight, after benchmarking, it is unlikely that we would want to set a limit greater than 2^31 - 1. - the number of checkpoints in an epoch, as the number of transactions in an epoch (a number that is guaranteed to be greater) is being represented using an `Int` at the moment (and soon a `UInt53`). This will be a breaking change, so will only go out with the new major version. Hopefully, this change will be minimal as the format of this scalar over the wire is the same as for `Int`, but it will require existing clients to register a new scalar in most cases. ## Test plan Existing tests: ``` sui-graphql-rpc$ cargo nextest run sui-graphql-e2e-tests$ cargo nextest run --features pg_integration ``` --- ## Release notes Check each box that your changes affect. If none of the boxes relate to your changes, release notes aren't required. For each box you select, include information after the relevant heading that describes the impact of your changes that a user might notice and any actions they must take to implement updates. - [ ] Protocol: - [ ] Nodes (Validators and Full nodes): - [ ] Indexer: - [ ] JSON-RPC: - [x] GraphQL: Introduces a new scalar -- `UInt53` -- to represent unsigned 53 bit integer values. Some uses of `Int` in the existing schema have been replaced with `UInt53`. All clients will need to register the new scalar and clients for statically typed languages will also need to use a wider (e.g. 64 bit), unsigned type to hold the value. - [ ] CLI: - [ ] Rust SDK:
tx-tomcat
pushed a commit
to tx-tomcat/sui-network
that referenced
this pull request
Jul 29, 2024
## Description Whilst working on MystenLabs#18337, I noticed that we were over-using the `Int` scalar -- using it to represent values that could exceed 2^31 - 1 -- when the GraphQL spec states that `Int` must be a 32-bit signed integer. We made this decision at the time (a) because `async-graphql` allowed converting `u64`s to `Int` and we were primarily concerned with the fact that although JSON doesn't specify a precision for its numeric types, JS (among other languages), assumes it is an IEEE double-precision floating point number, so can only represent integral values precisely below 2^53. `cynic` (a Rust GraphQL client library) is (correctly) stricter, however, and maps an `Int` to an `i32`, always. There may be other similarly strict client libraries for other languages. This PR introduces a new scalar, `UInt`, that maps to a JSON number literal, just like `Int`, but allows us to ascribe our own meaning (in this case, it will be an unsigned number, and it can be as large as 2^53). This scalar has been used in many cases where we had previously used `Int`: sequence numbers, counts of objects, checkpoints, transactions, etc. While other uses continue to use `Int` (pagination limits, service limits, values bounded by the number of validators). In some cases, we have switched from `BigInt` to using this scalar notably: - the db cost estimate, which was previously a `BigInt` because we were unsure of its scale, but in hindsight, after benchmarking, it is unlikely that we would want to set a limit greater than 2^31 - 1. - the number of checkpoints in an epoch, as the number of transactions in an epoch (a number that is guaranteed to be greater) is being represented using an `Int` at the moment (and soon a `UInt53`). This will be a breaking change, so will only go out with the new major version. Hopefully, this change will be minimal as the format of this scalar over the wire is the same as for `Int`, but it will require existing clients to register a new scalar in most cases. ## Test plan Existing tests: ``` sui-graphql-rpc$ cargo nextest run sui-graphql-e2e-tests$ cargo nextest run --features pg_integration ``` --- ## Release notes Check each box that your changes affect. If none of the boxes relate to your changes, release notes aren't required. For each box you select, include information after the relevant heading that describes the impact of your changes that a user might notice and any actions they must take to implement updates. - [ ] Protocol: - [ ] Nodes (Validators and Full nodes): - [ ] Indexer: - [ ] JSON-RPC: - [x] GraphQL: Introduces a new scalar -- `UInt53` -- to represent unsigned 53 bit integer values. Some uses of `Int` in the existing schema have been replaced with `UInt53`. All clients will need to register the new scalar and clients for statically typed languages will also need to use a wider (e.g. 64 bit), unsigned type to hold the value. - [ ] CLI: - [ ] Rust SDK:
amnn
force-pushed
the
amnn/gql-pkg-dump
branch
from
1c4097f to
9080e54
Compare
amnn
force-pushed
the
amnn/gql-pkg-dump
branch
from
9080e54 to
5af07d1
Compare
amnn
force-pushed
the
amnn/gql-pkg-dump
branch
from
5af07d1 to
9cf0651
Compare
amnn
force-pushed
the
amnn/gql-pkg-dump
branch
from
9cf0651 to
c52db3a
Compare
amnn
force-pushed
the
amnn/gql-pkg-dump
branch
from
c52db3a to
1411550
Compare
## Description Add a command for generating a `config.toml` file for the GraphQL service with all its parameters set to their default values. ## Test plan ``` cargo run --bin sui-graphql-rpc -- generate-config /tmp/config.toml ```
## Description Replace the original implementation of the dump-packages command (which requires access to an indexer database) with an implementation that reads from a GraphQL service. The former is not readily accessible, but the latter should be. The new tool is also able to run incrementally: Fetching only packages created before a certain checkpoint, or pick up where a previous invocation took off to fetch new packages that were introduced since. ## Test plan Ran a test invocation, on our experimental read replica. With a max page size of 200, I was able to fetch 17000 packages (all the packages at the time the read replica was created) in 3 minutes.
amnn
force-pushed
the
amnn/gql-pkg-dump
branch
from
1411550 to
657e435
Compare
amnn
added a commit
that referenced
this pull request
Aug 20, 2024
## Description Replace the original implementation of the dump-packages command (which requires access to an indexer database) with an implementation that reads from a GraphQL service. The former is not readily accessible, but the latter should be. The new tool is also able to run incrementally: Fetching only packages created before a certain checkpoint, or pick up where a previous invocation took off to fetch new packages that were introduced since. ## Test plan Ran a test invocation, on our experimental read replica. With a max page size of 200, I was able to fetch 17000 packages (all the packages at the time the read replica was created) in 3 minutes. ## Stack - #17543 - #17692 - #17693 - #17696 - #18287 - #18288 - #18336 --- ## Release notes Check each box that your changes affect. If none of the boxes relate to your changes, release notes aren't required. For each box you select, include information after the relevant heading that describes the impact of your changes that a user might notice and any actions they must take to implement updates. - [ ] Protocol: - [ ] Nodes (Validators and Full nodes): - [ ] Indexer: - [ ] JSON-RPC: - [ ] GraphQL: - [ ] CLI: - [ ] Rust SDK:
amnn
added a commit
that referenced
this pull request
Aug 20, 2024
## Description Replace the original implementation of the dump-packages command (which requires access to an indexer database) with an implementation that reads from a GraphQL service. The former is not readily accessible, but the latter should be. The new tool is also able to run incrementally: Fetching only packages created before a certain checkpoint, or pick up where a previous invocation took off to fetch new packages that were introduced since. ## Test plan Ran a test invocation, on our experimental read replica. With a max page size of 200, I was able to fetch 17000 packages (all the packages at the time the read replica was created) in 3 minutes. ## Stack - #17543 - #17692 - #17693 - #17696 - #18287 - #18288 - #18336 --- ## Release notes Check each box that your changes affect. If none of the boxes relate to your changes, release notes aren't required. For each box you select, include information after the relevant heading that describes the impact of your changes that a user might notice and any actions they must take to implement updates. - [ ] Protocol: - [ ] Nodes (Validators and Full nodes): - [ ] Indexer: - [ ] JSON-RPC: - [ ] GraphQL: - [ ] CLI: - [ ] Rust SDK:
tx-tomcat
pushed a commit
to tx-tomcat/sui-network
that referenced
this pull request
Aug 27, 2024
## Description Replace the original implementation of the dump-packages command (which requires access to an indexer database) with an implementation that reads from a GraphQL service. The former is not readily accessible, but the latter should be. The new tool is also able to run incrementally: Fetching only packages created before a certain checkpoint, or pick up where a previous invocation took off to fetch new packages that were introduced since. ## Test plan Ran a test invocation, on our experimental read replica. With a max page size of 200, I was able to fetch 17000 packages (all the packages at the time the read replica was created) in 3 minutes. ## Stack - MystenLabs#17543 - MystenLabs#17692 - MystenLabs#17693 - MystenLabs#17696 - MystenLabs#18287 - MystenLabs#18288 - MystenLabs#18336 --- ## Release notes Check each box that your changes affect. If none of the boxes relate to your changes, release notes aren't required. For each box you select, include information after the relevant heading that describes the impact of your changes that a user might notice and any actions they must take to implement updates. - [ ] Protocol: - [ ] Nodes (Validators and Full nodes): - [ ] Indexer: - [ ] JSON-RPC: - [ ] GraphQL: - [ ] CLI: - [ ] Rust SDK:
suiwombat
pushed a commit
that referenced
this pull request
Sep 16, 2024
## Description Replace the original implementation of the dump-packages command (which requires access to an indexer database) with an implementation that reads from a GraphQL service. The former is not readily accessible, but the latter should be. The new tool is also able to run incrementally: Fetching only packages created before a certain checkpoint, or pick up where a previous invocation took off to fetch new packages that were introduced since. ## Test plan Ran a test invocation, on our experimental read replica. With a max page size of 200, I was able to fetch 17000 packages (all the packages at the time the read replica was created) in 3 minutes. ## Stack - #17543 - #17692 - #17693 - #17696 - #18287 - #18288 - #18336 --- ## Release notes Check each box that your changes affect. If none of the boxes relate to your changes, release notes aren't required. For each box you select, include information after the relevant heading that describes the impact of your changes that a user might notice and any actions they must take to implement updates. - [ ] Protocol: - [ ] Nodes (Validators and Full nodes): - [ ] Indexer: - [ ] JSON-RPC: - [ ] GraphQL: - [ ] CLI: - [ ] Rust SDK:
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.
Description
Replace the original implementation of the dump-packages command (which requires access to an indexer database) with an implementation that reads from a GraphQL service. The former is not readily accessible, but the latter should be.
The new tool is also able to run incrementally: Fetching only packages created before a certain checkpoint, or pick up where a previous invocation took off to fetch new packages that were introduced since.
Test plan
Ran a test invocation, on our experimental read replica. With a max page size of 200, I was able to fetch 17000 packages (all the packages at the time the read replica was created) in 3 minutes.
Stack
objects_versiontable. #17543generate-configsub-command #18336Release notes
Check each box that your changes affect. If none of the boxes relate to your changes, release notes aren't required.
For each box you select, include information after the relevant heading that describes the impact of your changes that a user might notice and any actions they must take to implement updates.