Store: Make initial sync more robust

Added re-try mechanism for store inital sync, where if the initial sync fails, it tries to do the initial sync again for given timeout duration.

Signed-off-by: Kartik-Garg <kartik.garg@infracloud.io>

CHANGELOG.md

@@ -23,6 +23,7 @@ We use *breaking :warning:* to mark changes that are not backward compatible (re
 
 - [#5995](https://github.com/thanos-io/thanos/pull/5995) Sidecar: Loads the TLS certificate during startup.
 - [#6044](https://github.com/thanos-io/thanos/pull/6044) Receive: mark ouf of window errors as conflict, if out-of-window samples ingestion is activated
+- [#6050](https://github.com/thanos-io/thanos/pull/6050) Store: Re-try bucket store initial sync upon failure.
 - [#6066](https://github.com/thanos-io/thanos/pull/6066) Tracing: fixed panic because of nil sampler
 - [#6067](https://github.com/thanos-io/thanos/pull/6067) Receive: fixed panic when querying uninitialized TSDBs.
 

cmd/thanos/store.go

@@ -49,6 +49,11 @@ import (
 	"github.com/thanos-io/thanos/pkg/ui"
 )
 
+const (
+	timeoutDuration  = 30
+	intervalDuration = 10
+)
+
 type storeConfig struct {
 	indexCacheConfigs           extflag.PathOrContent
 	objStoreConfig              extflag.PathOrContent
@@ -381,14 +386,25 @@ func runStore(
 
 			level.Info(logger).Log("msg", "initializing bucket store")
 			begin := time.Now()
-			if err := bs.InitialSync(ctx); err != nil {
+
+			//This will stop retrying after set timeout duration.
+			initialSyncCtx, cancel := context.WithTimeout(context.Background(), timeoutDuration*time.Second)
+			defer cancel()
+
+			//Retry in case of error.
+			err := runutil.Retry(intervalDuration*time.Second, initialSyncCtx.Done(), func() error {
+				return bs.InitialSync(ctx)
+			})
+
+			if err != nil {
 				close(bucketStoreReady)
 				return errors.Wrap(err, "bucket store initial sync")
 			}
+
 			level.Info(logger).Log("msg", "bucket store ready", "init_duration", time.Since(begin).String())
 			close(bucketStoreReady)
 
-			err := runutil.Repeat(conf.syncInterval, ctx.Done(), func() error {
+			err = runutil.Repeat(conf.syncInterval, ctx.Done(), func() error {
 				if err := bs.SyncBlocks(ctx); err != nil {
 					level.Warn(logger).Log("msg", "syncing blocks failed", "err", err)
 				}