Rack Awareness, Sequence Policies, More complex policies and other Fixes #140
Conversation
…o longer maintained, it moves it into this library and changes it to use crossbeam-channel instead of plain-old crossbeam
| @@ -27,15 +27,4 @@ pub enum Concurrency { | |||
| /// extremely large batch sizes because each node can process the request immediately. The | |||
| /// downside is extra threads will need to be created (or takedn from a thread pool). | |||
| Parallel, | |||
|
|
|||
| /// Issue up to N commands in parallel threads. When a request completes, a new request | |||
There was a problem hiding this comment.
I just took the liberty to remove this completely in order to facilitate the sequence write patch (i.e. allowing batch reads to fast retry on a replica). @jonas32 's async patch offloads the responsibility for this stuff onto the runtime, so I thought it was a good tradeoff.
| @@ -26,7 +26,7 @@ rt-tokio = ["aerospike-rt/rt-tokio"] | |||
| rt-async-std = ["aerospike-rt/rt-async-std"] | |||
|
|
|||
| [dev-dependencies] | |||
| env_logger = "0.9.3" | |||
| env_logger = "0.9" | |||
There was a problem hiding this comment.
Allow other versions of env_logger means it can keep up to date.
| @@ -36,15 +36,15 @@ impl<'a> Partition<'a> { | |||
| } | |||
|
|
|||
| pub fn new_by_key(key: &'a Key) -> Self { | |||
| let mut rdr = Cursor::new(&key.digest[0..4]); | |||
| let mut rdr = Cursor::new(&key.digest[0..2]); | |||
There was a problem hiding this comment.
Made to be more similar to the C version. I can't remember why, I have a feeling that it did make a difference somehow, although it doesn't look like it.
CMoore-Darwinium
changed the title
| @@ -89,16 +89,14 @@ impl Queue { | |||
| ) | |||
| .await; | |||
|
|
|||
| if conn.is_err() { | |||
| let Ok(Ok(conn)) = conn else { | |||
There was a problem hiding this comment.
This is a bugfix to leaked connections. In the ? return case, it would not -- the counter.
| base64 = "0.11" | ||
| crossbeam-queue = "0.2" | ||
| rand = "0.7" | ||
| ripemd = "0.1" |
There was a problem hiding this comment.
ripemd160 was not maintained last time I checked and we were getting audit warnings
Hi @khaf , this is my attempt at a rebased pull request
It includes: