KokaKiwi/garage
1
0
Fork 0
forked from Deuxfleurs/garage
Code Issues Pull requests Projects Releases Wiki Activity

[reconnect-only-current] filter nodes to reconnect to

Browse source 
do not try reconnecting to nodes received from consul/kubernetes
discovery if they are not currently in the layout
This commit is contained in:
Alex 2024-02-16 10:50:41 +01:00
parent 7be3f15e45
commit e91576677e
Signed by untrusted user: lx
GPG key ID: 0E496D15096376BE
1 changed files with 16 additions and 5 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

21
src/rpc/system.rs
View file

@ -725,15 +725,18 @@ impl System {
async fn discovery_loop(self: &Arc<Self>, mut stop_signal: watch::Receiver<bool>) { async fn discovery_loop(self: &Arc<Self>, mut stop_signal: watch::Receiver<bool>) {
while !*stop_signal.borrow() { while !*stop_signal.borrow() {
let not_configured = self.ring.borrow().layout.check().is_err(); let n_connected = self
let no_peers = self.peering.get_peer_list().len() < self.replication_factor;
let expected_n_nodes = self.ring.borrow().layout.num_nodes();
let bad_peers = self
.peering .peering
.get_peer_list() .get_peer_list()
.iter() .iter()
.filter(|p| p.is_up()) .filter(|p| p.is_up())
.count() != expected_n_nodes; .count();
let not_configured = self.ring.borrow().layout.check().is_err();
let no_peers = n_connected < self.replication_factor;
let expected_n_nodes = self.ring.borrow().layout.num_nodes();
let bad_peers = n_connected != expected_n_nodes;
if not_configured || no_peers || bad_peers { if not_configured || no_peers || bad_peers {
info!("Doing a bootstrap/discovery step (not_configured: {}, no_peers: {}, bad_peers: {})", not_configured, no_peers, bad_peers); info!("Doing a bootstrap/discovery step (not_configured: {}, no_peers: {}, bad_peers: {})", not_configured, no_peers, bad_peers);
@ -780,6 +783,14 @@ impl System {
} }
} }
if !not_configured && !no_peers {
// If the layout is configured, and we already have some connections
// to other nodes in the cluster, we can skip trying to connect to
// nodes that are not in the cluster layout.
let ring = self.ring.borrow();
ping_list.retain(|(id, _)| ring.layout.node_ids().contains(&(*id).into()));
}
for (node_id, node_addr) in ping_list { for (node_id, node_addr) in ping_list {
let self2 = self.clone(); let self2 = self.clone();
tokio::spawn(async move { tokio::spawn(async move {