Consistent crash with Redis 6.2.4 #1124
Comments
nihohit
added a commit
to amazon-contributing/redis-rs
that referenced
this issue
Oct 26, 2023
This happens due to a consistent crash - RedisJSON/RedisJSON#1124
|
Since v2.6 GA the minimal supported Redis version is 7.2 |
nihohit
added a commit
to amazon-contributing/redis-rs
that referenced
this issue
Oct 26, 2023
This happens due to a consistent crash - RedisJSON/RedisJSON#1124
nihohit
added a commit
to amazon-contributing/redis-rs
that referenced
this issue
Oct 26, 2023
This happens due to a consistent crash - RedisJSON/RedisJSON#1124
nihohit
added a commit
to redis-rs/redis-rs
that referenced
this issue
Oct 31, 2023
This happens due to a consistent crash - RedisJSON/RedisJSON#1124
barshaul
pushed a commit
to amazon-contributing/redis-rs
that referenced
this issue
Dec 7, 2023
* Disable JSON module tests for redis 6.2.4. (redis-rs#980) This happens due to a consistent crash - RedisJSON/RedisJSON#1124 * Implement `FromRedisValue` for `Box<[T]>` and `Arc<[T]>` (redis-rs#799) * Update minimal rust version to 1.6.5 (redis-rs#982) It's required by regex v1.10.2. https://github.com/redis-rs/redis-rs/actions/runs/6760188489/job/18373652560?pr=981 * Support Mutual TLS (redis-rs#858) Add support for mutual TLS. Enable specifying client certificate and key and (optional) root certificate. * changing timeouts from usize and isize to f64 (redis-rs#988) Co-authored-by: Eythor Helgason <eytorhelgason@gmail.com> * Accept iterator at `ClusterClient` initialization (redis-rs#987) * Prefer routing to primary in a transaction. (redis-rs#986) * Rename route_pipeline to route_for_pipeline. There are 2 `route_pipeline` functions in the file, so we change one to avoid confusion. * Prefer routing to primary in a transaction. * CrossSlot error on pipeline with many routes. * Sync Pub/Sub - cache received pub/sub messages. (redis-rs#910) When sending subscibre/unsubscribe messages to the server, the pub/sub client might receive messages that it can mistake to be the response, and not return to the next `get_message` call. This fix ensures that pub/sub messages will be queued until the next `get_message` call, so that no message will be missed. * Fix sync cluster behavior with transactions. (redis-rs#983) The leading MULTI prevents the transaction from being properly routed. In order to route the request correctly, we drop the leading MULTI from the routable. * Release redis 0.23.4 / redis-test 0.2.4 (redis-rs#992) * Tests: Add retries to test cluster creation (redis-rs#994) * Revert redis-test versioning changes (redis-rs#993) Publishing to crates.io without the version specification is currently not possible. Reverting these changes for now until we have a better solution/workaround. * Fix features for `load_native_certs`. (redis-rs#996) The directive now matches the directives wrapping `load_native_certs`'s `use` statements. This would fail compilation for anyone accidentally using the "tls" feature with rustls. * Bump aHash to v0.8.6 (redis-rs#966) * Update Command expiration values to be an appropriate type (redis-rs#589) This change updates expiration values provided to some commands in the `Command` trait to be of type `i64` or `u64` where appropriate, replacing the previously used `usize` type. More specifically, the commands that are updated are - SETEX, PSETEX - Expiration value changed from `usize` to `u64` - EXPIRE, PEXPIRE, EXPIREAT, PEXPIREAT - Expiration value changed from `usize` to `i64` For the `*SETEX` commands, there is no mention of supporting negative values. And indeed, providing a negative value through redis-cli returns an error message. For the `*EXPIRE*` commands, I couldn't find the paragraph in the docs referenced in redis-rs#575, nor any other mention that the commands accept a negative expiration argument. However, in testing with redis-cli it seems a negative expiration value is valid and in checking the code I found it to confirm these observations. See [here][code1] and [here][code2]. Note that this is a breaking change. [code1]: https://github.com/redis/redis/blob/e3ef73dc2a557232c60c732705e8e6ff2050eba9/src/expire.c#L570-L571 [code2]: https://github.com/redis/redis/blob/e3ef73dc2a557232c60c732705e8e6ff2050eba9/src/expire.c#L483-L491 * Fix StreamId::contains_key signature (redis-rs#783) * Order in usage of ClusterParams. (redis-rs#997) * Move `tls_params` into `ClusterParams`. The two objects are used together, so it makes sense to join them. * Cluster: Contain `ClusterParams` internally. This matches the async cluster's structure, and ensures that fields that are added to `ClusterParams` are automatically added to the sync cluster * Release redis 0.24.0 / redis-test 0.3.0 (redis-rs#998) --------- Co-authored-by: Jacob O'Toole <jacob@jotpot.co.uk> Co-authored-by: spaceangel <129979014+sp-angel@users.noreply.github.com> Co-authored-by: eythorhel19 <70693645+eythorhel19@users.noreply.github.com> Co-authored-by: Eythor Helgason <eytorhelgason@gmail.com> Co-authored-by: Ruan Petterson <ruan@petterson.eng.br> Co-authored-by: James Lucas <jaymell@users.noreply.github.com> Co-authored-by: aumetra <aumetra@cryptolab.net> Co-authored-by: Josh Leeb-du Toit <mail@joshleeb.com> Co-authored-by: Ayush <ayushsingh1325@gmail.com>
@LiorKogan |
|
I wouldn't recommend using a milestone or a release candidate version in production. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Redis 6.2.4 crashes consistently when loaded with a release build of the current (90ba302) commit. This happens before the server finishes loading, with the command
redis-server --loadmodule /home/ubuntu/repositories/RedisJSON/target/release/librejson.so --logfile /tmp/redisgMYd0p/redis.log --port 40505 --bind 127.0.0.1.OS :
Ubuntu 22.0
architecture: x86-64
logs:
The text was updated successfully, but these errors were encountered: