add API to specify metadata request timeout #353
Conversation
wprzytula
changed the title
There was a problem hiding this comment.
Pull Request Overview
This PR implements a new API function to allow users to configure server-side timeouts for metadata queries in the Cassandra driver. This addresses issues with large schemas where the default timeout may be too restrictive.
- Adds
cass_cluster_set_metadata_request_serverside_timeoutfunction to configure metadata query timeouts - Exposes existing Rust driver functionality through the C API
- Sets default timeout to 2 seconds with option to use cluster default by passing 0
Reviewed Changes
Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.
| File | Description |
|---|---|
include/cassandra.h |
Adds public API declaration with documentation for the new timeout function |
scylla-rust-wrapper/src/cluster.rs |
Implements the C function that sets metadata request timeout on the cluster configuration |
scylla-rust-wrapper/src/api.rs |
Exports the new function in the public API module |
Comments suppressed due to low confidence (1)
wprzytula
force-pushed
the
expose-api-for-metadata-request-timeouts
branch
2 times, most recently
from
6cf974a to
f3581c5
Compare
| /** | ||
| * Sets the server-side timeout for metadata queries. | ||
| * | ||
| * It means that all metadata queries will be set the given timeout | ||
| * no matter what timeout is set as a cluster default. | ||
| * This prevents timeouts of schema queries when the schema is large | ||
| * and the default timeout is configured as tight. | ||
| * | ||
| * <b>Default:</b> 2 seconds. | ||
| * | ||
| * @public @memberof CassCluster | ||
| * | ||
| * @param[in] cluster | ||
| * @param[in] timeout_ms Request timeout in milliseconds. Pass 0 to use cluster default timeout. | ||
| */ | ||
| CASS_EXPORT void | ||
| cass_cluster_set_metadata_request_serverside_timeout(CassCluster* cluster, | ||
| cass_duration_t timeout_ms); | ||
|
|
||
|
|
There was a problem hiding this comment.
Is there a dedicated setting for metadata client-side timeout?
If so, can we make it that:
- Default for server-side timeout is client-side timeout (perhaps with a slight reduction)
- There is a warning if users sets server-side higher than client-side?
If not, what client-side timeout is used? Can the above points apply to it?
There was a problem hiding this comment.
Is there a dedicated setting for metadata client-side timeout?
No.
There was a problem hiding this comment.
If not, what client-side timeout is used?
As with the other settings configured on execution profiles and on statements directly, the hierarchy is as follows:
statement > execution profile > cluster
There was a problem hiding this comment.
Can the above points apply to it?
I'm not sure I understand.
This commit adds a new API function to the driver: `cass_cluster_set_metadata_request_serverside_timeout`. This function allows users to set a server-side timeout for metadata queries, which is particularly useful for large schemas where the default timeout may be too tight. The default timeout is set to 2 seconds (as in Rust Driver), and users can pass 0 to use the cluster's default timeout. This simply exposes the existing Rust Driver's API. Fixes: scylladb#172
wprzytula
force-pushed
the
expose-api-for-metadata-request-timeouts
branch
from
f3581c5 to
59a11f6
Compare
This adds a new API function to the driver:
cass_cluster_set_metadata_request_serverside_timeout. This function allows users to set a server-side timeout for metadata queries, which is particularly useful for large schemas where the default timeout may be too tight. The default timeout is set to 2 seconds (as in Rust Driver), and users can pass 0 to use the cluster's default timeout. This simply exposes the existing Rust Driver's API.Fixes: #172
Pre-review checklist
[ ] I have implemented Rust unit tests for the features/changes introduced.[ ] I have enabled appropriate tests inMakefilein{SCYLLA,CASSANDRA}_(NO_VALGRIND_)TEST_FILTER.Fixes:annotations to PR description.