[#1680] replicator: Consider nodes under maintenance as OK

Nodes under maintenance SHOULD not respond to object requests. Based on
this, storage node's Policer SHOULD consider such nodes as problem ones.
However, to prevent spam with the new replicas, on the contrary, Policer
should consider them normal.

Make `Policer.processNodes` to exclude elements if `IsMaintenance()`
with shortage decreasing.

Signed-off-by: Leonard Lyubich <ctulhurider@gmail.com>
This commit is contained in:
Leonard Lyubich 2022-10-07 13:33:37 +04:00 committed by fyrchik
parent df5d7bf729
commit e99e25b52f
2 changed files with 16 additions and 0 deletions

View file

@ -23,6 +23,7 @@ Changelog for NeoFS Node
- Allow to evacuate shard data with `EvacuateShard` control RPC (#1800) - Allow to evacuate shard data with `EvacuateShard` control RPC (#1800)
- Flush write-cache when moving shard to DEGRADED mode (#1825) - Flush write-cache when moving shard to DEGRADED mode (#1825)
- Make `morph.cache_ttl` default value equal to morph block time (#1846) - Make `morph.cache_ttl` default value equal to morph block time (#1846)
- Policer marks nodes under maintenance as OK without requests (#1680)
### Fixed ### Fixed
- Description of command `netmap nodeinfo` (#1821) - Description of command `netmap nodeinfo` (#1821)

View file

@ -2,6 +2,7 @@ package policer
import ( import (
"context" "context"
"encoding/hex"
"github.com/nspcc-dev/neofs-node/pkg/core/container" "github.com/nspcc-dev/neofs-node/pkg/core/container"
"github.com/nspcc-dev/neofs-node/pkg/local_object_storage/engine" "github.com/nspcc-dev/neofs-node/pkg/local_object_storage/engine"
@ -142,6 +143,18 @@ func (p *Policer) processNodes(ctx *processPlacementContext, addr oid.Address,
nodes []netmap.NodeInfo, shortage uint32, checkedNodes *nodeCache) { nodes []netmap.NodeInfo, shortage uint32, checkedNodes *nodeCache) {
prm := new(headsvc.RemoteHeadPrm).WithObjectAddress(addr) prm := new(headsvc.RemoteHeadPrm).WithObjectAddress(addr)
handleMaintenance := func(node netmap.NodeInfo) {
// consider remote nodes under maintenance as problem OK. Such
// nodes MAY not respond with object, however, this is how we
// prevent spam with new replicas.
checkedNodes.submitReplicaHolder(node)
shortage--
p.log.Debug("consider node under maintenance as OK",
zap.String("node", hex.EncodeToString(node.PublicKey())),
)
}
for i := 0; shortage > 0 && i < len(nodes); i++ { for i := 0; shortage > 0 && i < len(nodes); i++ {
select { select {
case <-ctx.Done(): case <-ctx.Done():
@ -153,6 +166,8 @@ func (p *Policer) processNodes(ctx *processPlacementContext, addr oid.Address,
ctx.needLocalCopy = true ctx.needLocalCopy = true
shortage-- shortage--
} else if nodes[i].IsMaintenance() {
handleMaintenance(nodes[i])
} else { } else {
if status := checkedNodes.processStatus(nodes[i]); status >= 0 { if status := checkedNodes.processStatus(nodes[i]); status >= 0 {
if status == 0 { if status == 0 {