Files
seaweedfs/weed/operation/lookup_vid_cache_test.go
Chris Lu cf64cafc3b volume: drop stale volume-location cache on under-replication (#10185)
* volume: drop stale volume-location cache on under-replication

A replicated write looks up the volume's locations and caches them for 10
minutes. When the master briefly reports fewer replicas than the copy count
(e.g. a stale heartbeat drops a just-added volume), that under-replicated
result got cached, so every write failed with "replicating operations is less
than replication copy count" until the entry expired -- long after the master
re-registered the replica.

Invalidate the cached entry when the location count is below the copy count, so
the next write re-queries the master and recovers as soon as it heals.

* volume: mirror the replication copy-count guard in seaweed-volume

do_replicated_request accepted a write even when the master reported fewer
locations than the volume's copy count, silently under-replicating. Reject it,
matching Go's GetWritableRemoteReplications. lookup_volume is uncached, so the
next write recovers as soon as the missing replica re-registers.
2026-07-01 13:51:59 -07:00

66 lines
1.7 KiB
Go

package operation
import (
"fmt"
"testing"
"time"
)
func TestCaching(t *testing.T) {
var (
vc VidCache
)
var locations []Location
locations = append(locations, Location{Url: "a.com:8080"})
vc.Set("123", locations, time.Second)
ret, _ := vc.Get("123")
if ret == nil {
t.Fatal("Not found vid 123")
}
fmt.Printf("vid 123 locations = %v\n", ret)
time.Sleep(2 * time.Second)
ret, _ = vc.Get("123")
if ret != nil {
t.Fatal("Not found vid 123")
}
}
// a stale under-replicated result must not linger for the full cache TTL
func TestCachingDelete(t *testing.T) {
var vc VidCache
locations := []Location{{Url: "a.com:8080"}}
vc.Set("123", locations, time.Minute)
if ret, _ := vc.Get("123"); ret == nil {
t.Fatal("expected vid 123 to be cached")
}
vc.Delete("123")
if ret, _ := vc.Get("123"); ret != nil {
t.Fatal("expected vid 123 to be evicted after Delete")
}
// deleting a missing or out-of-range id must not panic
vc.Delete("123")
vc.Delete("4294967296")
}
// a single large volume id must not allocate an entry per id below it
func TestCachingLargeVolumeId(t *testing.T) {
var vc VidCache
locations := []Location{{Url: "a.com:8080"}}
vc.Set("32000000", locations, time.Minute)
if got := len(vc.cache); got != 1 {
t.Fatalf("expected 1 cached entry, got %d", got)
}
if ret, _ := vc.Get("32000000"); ret == nil {
t.Fatal("Not found vid 32000000")
}
// ids beyond uint32 are not real volume ids and must not wrap into the cache
vc.Set("4294967296", locations, time.Minute)
if got := len(vc.cache); got != 1 {
t.Fatalf("out-of-range id must not be cached, got %d entries", got)
}
if ret, _ := vc.Get("4294967296"); ret != nil {
t.Fatal("out-of-range vid 4294967296 should not be found")
}
}