Fix typo

cachelib/cachebench/cache/ItemRecords.h

@@ -54,11 +54,11 @@ namespace facebook::cachelib::cachebench {
  * No matter how the state of the item is changed in cache, in RAM, NVM,
  * evicted, or removed, the record will be kept in ItemRecords until the end of
  * test.
- * The version field is the latest version of the item (via in-palce or
- * chianed mutation). This ensures that the destructor is only triggers for
+ * The version field is the latest version of the item (via in-place or
+ * chained mutation). This ensures that the destructor is only triggered for
  * the latest version of the item.
  * The destructCount field is incremented in the ItemDestructor, it ensures that
- * each allocation triggers destructor once and only once.
+ * each allocation triggers the destructor once and only once.
  */
 struct ItemRecord {
   uint32_t destructCount{0};

cachelib/persistence/tests/PersistenceManagerTest.cpp

@@ -94,7 +94,7 @@ TEST_F(PersistenceManagerTest, testConfigChange) {
 
 TEST_F(PersistenceManagerTest, testSmallSinglePool) {
   std::string key = "key";
-  std::string data = "Repalce the data associated with key key";
+  std::string data = "Replace the data associated with key key";
   // test single item, one pool
   cache_.test({{key, data}}, 1, 0, false);
 }

website/docs/Cache_Library_User_Guides/Configuring_cachebench_parameters.md

@@ -115,9 +115,9 @@ cachebench supports generating synthetic workloads using a few techniques. The t
 Generates popularity of keys through a discrete distribution specified in  *popularityBuckets* and *popularityWeights* parameter. Discrete sizes are generated through a discrete distribution specified through `valSizeRange` and `valSizeRangeProbability`. The value size configuration can be provided inline as an array or through a `valSizeDistFile` in json format.
 
 * **normal**
-Uses normal workload distribution for popularity of keys as opposed to discrete popularity buckets. For value sizes, it supports both discrete and continuous value size distribution. To use discrete value size distribution, the `valSizeRangeProbabilithy` should have same number of values as `valSizeRange` array. When `valSizeRangeProbabilithy` contains one less member than `valSizeRange`, we interpret the probability as corresponding to each interval in `valSizeRange` and use a piecewise_constant_distribution.
+Uses normal workload distribution for popularity of keys as opposed to discrete popularity buckets. For value sizes, it supports both discrete and continuous value size distribution. To use discrete value size distribution, the `valSizeRangeProbability` should have same number of values as `valSizeRange` array. When `valSizeRangeProbability` contains one less member than `valSizeRange`, we interpret the probability as corresponding to each interval in `valSizeRange` and use a piecewise_constant_distribution.
 
-In all above setups, cachebench overrides the `valSizeRange` and `vaSizeRangeProbability` from inline json array if `valSizeDistFile` is present.
+In all above setups, cachebench overrides the `valSizeRange` and `valSizeRangeProbability` from inline json array if `valSizeDistFile` is present.
 
 ### Throttling the benchmark
 

website/docs/Cache_Library_User_Guides/Write_data_to_cache.md

@@ -168,7 +168,7 @@ std::memcpy(handle->getMemory(), data.data(), data.size());
 // Insert the item handle into the cache.
 cache->insertOrReplace(handle);
 
-data = "Repalce the data associated with key key1";
+data = "Replace the data associated with key key1";
 // Allocate memory for the replacement data.
 handle = cache->allocate(pool_id, "key1", data.size());