diff --git a/app/vmselect/promql/rollup_result_cache.go b/app/vmselect/promql/rollup_result_cache.go
index 201c18f4f..63b2cc96c 100644
--- a/app/vmselect/promql/rollup_result_cache.go
+++ b/app/vmselect/promql/rollup_result_cache.go
@@ -441,7 +441,7 @@ func mustSaveRollupResultCacheKeyPrefix(path string) {
 var tooBigRollupResults = metrics.NewCounter("vm_too_big_rollup_results_total")
 
 // Increment this value every time the format of the cache changes.
-const rollupResultCacheVersion = 8
+const rollupResultCacheVersion = 9
 
 func marshalRollupResultCacheKey(dst []byte, expr metricsql.Expr, window, step int64, etfs [][]storage.TagFilter) []byte {
 	dst = append(dst, rollupResultCacheVersion)
diff --git a/docs/CHANGELOG.md b/docs/CHANGELOG.md
index af437cc40..493a3fd49 100644
--- a/docs/CHANGELOG.md
+++ b/docs/CHANGELOG.md
@@ -15,10 +15,20 @@ The following tip changes can be tested by building VictoriaMetrics components f
 
 ## tip
 
+## [v1.89.1](https://github.com/VictoriaMetrics/VictoriaMetrics/releases/tag/v1.89.1)
+
+Released at 2023-03-12
+
+* BUGFIX: prevent from possible `cannot unmarshal timeseries from rollupResultCache` panic after the upgrade to [v1.89.0](https://docs.victoriametrics.com/CHANGELOG.html#v1890).
+
 ## [v1.89.0](https://github.com/VictoriaMetrics/VictoriaMetrics/releases/tag/v1.89.0)
 
 Released at 2023-03-12
 
+**Update note: this release can crash with `cannot unmarshal timeseries from rollupResultCache` panic after the upgrade from the previous releases.
+This issue can be fixed by removing caches stored on disk according to [these docs](https://docs.victoriametrics.com/#cache-removal).
+Another option is to upgrade to [v1.89.1](https://docs.victoriametrics.com/CHANGELOG.html#v1891).**
+
 * SECURITY: upgrade Go builder from Go1.20.1 to Go1.20.2. See [the list of issues addressed in Go1.20.2](https://github.com/golang/go/issues?q=milestone%3AGo1.20.2+label%3ACherryPickApproved).
 
 * FEATURE: [vmctl](https://docs.victoriametrics.com/vmctl.html): increase the default value for `--remote-read-http-timeout` command-line option from 30s (30 seconds) to 5m (5 minutes). This reduces the probability of timeout errors when migrating big number of time series. See [this pull request](https://github.com/VictoriaMetrics/VictoriaMetrics/pull/3879).