github/kata-containers
1
0
Fork 1
mirror of https://github.com/kata-containers/kata-containers.git synced 2025-06-28 08:17:37 +00:00
Code Issues Projects Releases Wiki Activity

agent: add retry between doing CPU hotplug and make it online.

Browse Source 
Sometimes runtime will fail in onlining CPU process,
because when the runtime calls to QMP
`device_add`, QEMU doesn't allocate all vCPUs inmediatelly.

Fixes: #665

Signed-off-by: bin liu <bin@hyper.sh>
This commit is contained in:
bin liu 2020-09-09 23:09:37 +08:00
parent 7f20587433
commit d86e74674c
2 changed files with 41 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
src/agent/src/rpc.rs
View File

@ -1179,8 +1179,6 @@ impl protocols::agent_ttrpc::AgentService for agentService {
_ctx: &ttrpc::TtrpcContext, _ctx: &ttrpc::TtrpcContext,
req: protocols::agent::OnlineCPUMemRequest, req: protocols::agent::OnlineCPUMemRequest,
) -> ttrpc::Result<Empty> { ) -> ttrpc::Result<Empty> {
// sleep 5 seconds for debug
// thread::sleep(Duration::new(5, 0));
let s = Arc::clone(&self.sandbox); let s = Arc::clone(&self.sandbox);
let sandbox = s.lock().unwrap(); let sandbox = s.lock().unwrap();

44
src/agent/src/sandbox.rs
View File

@ -25,6 +25,7 @@ use std::fs;
use std::os::unix::fs::PermissionsExt; use std::os::unix::fs::PermissionsExt;
use std::path::Path; use std::path::Path;
use std::sync::mpsc::Sender; use std::sync::mpsc::Sender;
use std::{thread, time};
#[derive(Debug)] #[derive(Debug)]
pub struct Sandbox { pub struct Sandbox {
@ -316,10 +317,18 @@ fn online_resources(logger: &Logger, path: &str, pattern: &str, num: i32) -> Res
if re.is_match(name) { if re.is_match(name) {
let file = format!("{}/{}", p.to_str().unwrap(), SYSFS_ONLINE_FILE); let file = format!("{}/{}", p.to_str().unwrap(), SYSFS_ONLINE_FILE);
info!(logger, "{}", file.as_str()); info!(logger, "{}", file.as_str());
let c = fs::read_to_string(file.as_str())?;
let c = fs::read_to_string(file.as_str());
if c.is_err() {
continue;
}
let c = c.unwrap();
if c.trim().contains("0") { if c.trim().contains("0") {
fs::write(file.as_str(), "1")?; let r = fs::write(file.as_str(), "1");
if r.is_err() {
continue;
}
count += 1; count += 1;
if num > 0 && count == num { if num > 0 && count == num {
@ -336,8 +345,37 @@ fn online_resources(logger: &Logger, path: &str, pattern: &str, num: i32) -> Res
Ok(0) Ok(0)
} }
// max wait for all CPUs to online will use 50 * 100 = 5 seconds.
const ONLINE_CPUMEM_WATI_MILLIS: u64 = 50;
const ONLINE_CPUMEM_MAX_RETRIES: u32 = 100;
fn online_cpus(logger: &Logger, num: i32) -> Result<i32> { fn online_cpus(logger: &Logger, num: i32) -> Result<i32> {
online_resources(logger, SYSFS_CPU_ONLINE_PATH, r"cpu[0-9]+", num) let mut onlined_count: i32 = 0;
for i in 0..ONLINE_CPUMEM_MAX_RETRIES {
let r = online_resources(
logger,
SYSFS_CPU_ONLINE_PATH,
r"cpu[0-9]+",
(num - onlined_count),
);
if r.is_err() {
return r;
}
onlined_count += r.unwrap();
if onlined_count == num {
info!(logger, "online {} CPU(s) after {} retries", num, i);
return Ok(num);
}
thread::sleep(time::Duration::from_millis(ONLINE_CPUMEM_WATI_MILLIS));
}
Err(anyhow!(
"failed to online {} CPU(s) after {} retries",
num,
ONLINE_CPUMEM_MAX_RETRIES
))
} }
fn online_memory(logger: &Logger) -> Result<()> { fn online_memory(logger: &Logger) -> Result<()> {