This is an automated email from the ASF dual-hosted git repository.
agrove pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/arrow-ballista.git
The following commit(s) were added to refs/heads/master by this push:
new 9ad583ea Downgrade `docker-compose.yaml` to version 3.3 so that we can
support Ubuntu 20.04.4 LTS (#329)
9ad583ea is described below
commit 9ad583eaa9b4feae8b03c7c544eb8d3fa8b83a2d
Author: Andy Grove <[email protected]>
AuthorDate: Sat Oct 15 15:18:22 2022 -0600
Downgrade `docker-compose.yaml` to version 3.3 so that we can support
Ubuntu 20.04.4 LTS (#329)
* Use Docker version 3.3 so that we can support Ubuntu 20.04.4 LTS
* fix labeler
* executor retry connect to scheduler
* log info message when connected to scheduler
---
ballista/rust/executor/executor_config_spec.toml | 6 +++
ballista/rust/executor/src/main.rs | 47 +++++++++++++++++++++---
benchmarks/README.md | 1 +
docker-compose.yml | 8 ++--
4 files changed, 53 insertions(+), 9 deletions(-)
diff --git a/ballista/rust/executor/executor_config_spec.toml
b/ballista/rust/executor/executor_config_spec.toml
index f8f8c5a1..7b9f11dd 100644
--- a/ballista/rust/executor/executor_config_spec.toml
+++ b/ballista/rust/executor/executor_config_spec.toml
@@ -60,6 +60,12 @@ type = "u16"
default = "50052"
doc = "bind grpc service port"
+[[param]]
+name = "scheduler_connect_timeout_seconds"
+type = "u16"
+default = "0"
+doc = "How long to try connecting to scheduler before failing. Set to zero to
fail after first attempt."
+
[[param]]
name = "work_dir"
type = "String"
diff --git a/ballista/rust/executor/src/main.rs
b/ballista/rust/executor/src/main.rs
index 9688377a..260a7204 100644
--- a/ballista/rust/executor/src/main.rs
+++ b/ballista/rust/executor/src/main.rs
@@ -20,13 +20,13 @@
use chrono::{DateTime, Duration, Utc};
use std::net::SocketAddr;
use std::sync::Arc;
-use std::time::Duration as Core_Duration;
+use std::time::{Duration as Core_Duration, Instant};
use std::{env, io};
use anyhow::{Context, Result};
use arrow_flight::flight_service_server::FlightServiceServer;
use ballista_executor::{execution_loop, executor_server};
-use log::{error, info};
+use log::{error, info, warn};
use tempfile::TempDir;
use tokio::fs::ReadDir;
use tokio::signal;
@@ -181,9 +181,46 @@ async fn main() -> Result<()> {
concurrent_tasks,
));
- let connection = create_grpc_client_connection(scheduler_url)
- .await
- .context("Could not connect to scheduler")?;
+ let connect_timeout = opt.scheduler_connect_timeout_seconds as u64;
+ let connection = if connect_timeout == 0 {
+ create_grpc_client_connection(scheduler_url)
+ .await
+ .context("Could not connect to scheduler")
+ } else {
+ // this feature was added to support docker-compose so that we can
have the executor
+ // wait for the scheduler to start, or at least run for 10 seconds
before failing so
+ // that docker-compose's restart policy will restart the container.
+ let start_time = Instant::now().elapsed().as_secs();
+ let mut x = None;
+ while x.is_none()
+ && Instant::now().elapsed().as_secs() - start_time <
connect_timeout
+ {
+ match create_grpc_client_connection(scheduler_url.clone())
+ .await
+ .context("Could not connect to scheduler")
+ {
+ Ok(conn) => {
+ info!("Connected to scheduler at {}", scheduler_url);
+ x = Some(conn);
+ }
+ Err(e) => {
+ warn!(
+ "Failed to connect to scheduler at {} ({}); retrying
...",
+ scheduler_url, e
+ );
+ std::thread::sleep(time::Duration::from_millis(500));
+ }
+ }
+ }
+ match x {
+ Some(conn) => Ok(conn),
+ _ => Err(BallistaError::General(format!(
+ "Timed out attempting to connect to scheduler at {}",
+ scheduler_url
+ ))
+ .into()),
+ }
+ }?;
let mut scheduler = SchedulerGrpcClient::new(connection);
diff --git a/benchmarks/README.md b/benchmarks/README.md
index 6b0acbaa..11b35b93 100644
--- a/benchmarks/README.md
+++ b/benchmarks/README.md
@@ -151,6 +151,7 @@ cargo run --release --bin tpch benchmark ballista --host
localhost --port 50050
To start a Rust scheduler and executor using Docker Compose:
```bash
+cargo build --release
docker-compose up --build
```
diff --git a/docker-compose.yml b/docker-compose.yml
index 5e71f2e5..f2459236 100644
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -14,7 +14,7 @@
# KIND, either express or implied. See the License for the
# specific language governing permissions and limitations
# under the License.
-version: '3.8'
+version: '3.3'
services:
etcd:
image: quay.io/coreos/etcd:v3.4.9
@@ -46,16 +46,16 @@ services:
build:
dockerfile: dev/docker/ballista-executor.Dockerfile
context: .
- command: "--bind-host 0.0.0.0 --scheduler-host ballista-scheduler"
+ command: "--bind-host 0.0.0.0 --scheduler-host ballista-scheduler
--scheduler-connect-timeout-seconds 15"
deploy:
replicas: 2
+ restart: always
environment:
- RUST_LOG=ballista=debug,info
volumes:
- ./benchmarks/data:/data
depends_on:
- ballista-scheduler:
- condition: service_healthy
+ - ballista-scheduler
healthcheck:
test: ["CMD", "nc", "-z", "ballista-executor", "50051"]
interval: 5s
