Files
neon/pageserver/src/control_plane_client.rs
John Spray 61d661a6c3 pageserver: generation number fetch on startup and use in /attach (#5163)
## Problem

- #5050 

Closes: https://github.com/neondatabase/neon/issues/5136

## Summary of changes

- A new configuration property `control_plane_api` controls other
functionality in this PR: if it is unset (default) then everything still
works as it does today.
- If `control_plane_api` is set, then on startup we call out to control
plane `/re-attach` endpoint to discover our attachments and their
generations. If an attachment is missing from the response we implicitly
detach the tenant.
- Calls to pageserver `/attach` API may include a `generation`
parameter. If `control_plane_api` is set, then this parameter is
mandatory.
- RemoteTimelineClient's loading of index_part.json is generation-aware,
and will try to load the index_part with the most recent generation <=
its own generation.
- The `neon_local` testing environment now includes a new binary
`attachment_service` which implements the endpoints that the pageserver
requires to operate. This is on by default if running `cargo neon` by
hand. In `test_runner/` tests, it is off by default: existing tests
continue to run with in the legacy generation-less mode.

Caveats:
- The re-attachment during startup assumes that we are only re-attaching
tenants that have previously been attached, and not totally new tenants
-- this relies on the control plane's attachment logic to keep retrying
so that we should eventually see the attach API call. That's important
because the `/re-attach` API doesn't tell us which timelines we should
attach -- we still use local disk state for that. Ref:
https://github.com/neondatabase/neon/issues/5173
- Testing: generations are only enabled for one integration test right
now (test_pageserver_restart), as a smoke test that all the machinery
basically works. Writing fuller tests that stress tenant migration will
come later, and involve extending our test fixtures to deal with
multiple pageservers.
- I'm not in love with "attachment_service" as a name for the neon_local
component, but it's not very important because we can easily rename
these test bits whenever we want.
- Limited observability when in re-attach on startup: when I add
generation validation for deletions in a later PR, I want to wrap up the
control plane API calls in some small client class that will expose
metrics for things like errors calling the control plane API, which will
act as a strong red signal that something is not right.

Co-authored-by: Christian Schwarz <christian@neon.tech>
Co-authored-by: Joonas Koivunen <joonas@neon.tech>
2023-09-06 14:44:48 +01:00

120 lines
4.0 KiB
Rust

use std::collections::HashMap;
use hyper::StatusCode;
use pageserver_api::control_api::{ReAttachRequest, ReAttachResponse};
use tokio_util::sync::CancellationToken;
use url::Url;
use utils::{
backoff,
generation::Generation,
id::{NodeId, TenantId},
};
use crate::config::PageServerConf;
// Backoffs when control plane requests do not succeed: compromise between reducing load
// on control plane, and retrying frequently when we are blocked on a control plane
// response to make progress.
const BACKOFF_INCREMENT: f64 = 0.1;
const BACKOFF_MAX: f64 = 10.0;
/// The Pageserver's client for using the control plane API: this is a small subset
/// of the overall control plane API, for dealing with generations (see docs/rfcs/025-generation-numbers.md)
pub(crate) struct ControlPlaneClient {
http_client: reqwest::Client,
base_url: Url,
node_id: NodeId,
cancel: CancellationToken,
}
impl ControlPlaneClient {
/// A None return value indicates that the input `conf` object does not have control
/// plane API enabled.
pub(crate) fn new(conf: &'static PageServerConf, cancel: &CancellationToken) -> Option<Self> {
let mut url = match conf.control_plane_api.as_ref() {
Some(u) => u.clone(),
None => return None,
};
if let Ok(mut segs) = url.path_segments_mut() {
// This ensures that `url` ends with a slash if it doesn't already.
// That way, we can subsequently use join() to safely attach extra path elements.
segs.pop_if_empty().push("");
}
let client = reqwest::ClientBuilder::new()
.build()
.expect("Failed to construct http client");
Some(Self {
http_client: client,
base_url: url,
node_id: conf.id,
cancel: cancel.clone(),
})
}
async fn try_re_attach(
&self,
url: Url,
request: &ReAttachRequest,
) -> anyhow::Result<ReAttachResponse> {
match self.http_client.post(url).json(request).send().await {
Err(e) => Err(anyhow::Error::from(e)),
Ok(r) => {
if r.status() == StatusCode::OK {
r.json::<ReAttachResponse>()
.await
.map_err(anyhow::Error::from)
} else {
Err(anyhow::anyhow!("Unexpected status {}", r.status()))
}
}
}
}
/// Block until we get a successful response
pub(crate) async fn re_attach(&self) -> anyhow::Result<HashMap<TenantId, Generation>> {
let re_attach_path = self
.base_url
.join("re-attach")
.expect("Failed to build re-attach path");
let request = ReAttachRequest {
node_id: self.node_id,
};
let mut attempt = 0;
loop {
let result = self.try_re_attach(re_attach_path.clone(), &request).await;
match result {
Ok(res) => {
tracing::info!(
"Received re-attach response with {} tenants",
res.tenants.len()
);
return Ok(res
.tenants
.into_iter()
.map(|t| (t.id, Generation::new(t.generation)))
.collect::<HashMap<_, _>>());
}
Err(e) => {
tracing::error!("Error re-attaching tenants, retrying: {e:#}");
backoff::exponential_backoff(
attempt,
BACKOFF_INCREMENT,
BACKOFF_MAX,
&self.cancel,
)
.await;
if self.cancel.is_cancelled() {
return Err(anyhow::anyhow!("Shutting down"));
}
attempt += 1;
}
}
}
}
}