DOC-745 | Misc release notes 3.12.5 (#712)

* --server.maximal-number-sync-shard-actions release notes

* Changed RocksDB periodic compaction default

Author: Simran-B

site/content/3.12/release-notes/version-3.12/incompatible-changes-in-3-12.md

@@ -1054,7 +1054,7 @@ This slightly decreases performance due to using the block cache for additional
 things, and you may need to allow ArangoDB to use more memory for the RocksDB
 block cache than before with the `--rocksdb.block-cache-size` startup option.
 
-#### `mmap` log topic removed
+### `mmap` log topic removed
 
 <small>Introduced in: v3.12.1</small>
 
@@ -1063,6 +1063,18 @@ unused since v3.12.0 and has now been removed. Attempts to set the log level for
 this topic logs a warning, for example, using a startup option like
 `--log.level mmap=trace`.
 
+### Daily RocksDB background compactions
+
+<small>Introduced in: v3.12.5</small>
+
+The default value of the `--rocksdb.periodic-compaction-ttl` startup option has
+been changed from 30 days to 24 hours.
+
+Certain deleted keys ranges in RocksDB may only get compacted by periodic
+background compactions. The changed default ensures this cleanup to happen more
+frequently. Compactions can potentially lead to spikes in CPU, memory, and I/O
+usage. You may now observe this daily instead of monthly.
+
 ## Client tools
 
 ### arangodump

site/content/3.12/release-notes/version-3.12/whats-new-in-3-12.md

@@ -1688,6 +1688,18 @@ The following startup options for cluster deployments have been added:
   The delay (in seconds) before shutting down a Coordinator if no heartbeat can
   be sent. Set to `0` to deactivate this shutdown.
 
+### Limit for shard synchronization actions
+
+<small>Introduced in: v3.11.14, v3.12.5</small>
+
+The number of `SynchronizeShard` actions that can be scheduled internally by the
+cluster maintenance has been restricted to prevent these actions from blocking
+`TakeoverShardLeadership` actions with a higher priority, which could lead to
+service interruption during upgrades and after failovers.
+
+The new `--server.maximal-number-sync-shard-actions` startup option controls
+how many `SynchronizeShard` actions can be queued at any given time.
+
 ## Miscellaneous changes
 
 ### V8 and ICU library upgrades

site/content/3.13/release-notes/version-3.12/incompatible-changes-in-3-12.md

@@ -1054,7 +1054,7 @@ This slightly decreases performance due to using the block cache for additional
 things, and you may need to allow ArangoDB to use more memory for the RocksDB
 block cache than before with the `--rocksdb.block-cache-size` startup option.
 
-#### `mmap` log topic removed
+### `mmap` log topic removed
 
 <small>Introduced in: v3.12.1</small>
 
@@ -1063,6 +1063,18 @@ unused since v3.12.0 and has now been removed. Attempts to set the log level for
 this topic logs a warning, for example, using a startup option like
 `--log.level mmap=trace`.
 
+### Daily RocksDB background compactions
+
+<small>Introduced in: v3.12.5</small>
+
+The default value of the `--rocksdb.periodic-compaction-ttl` startup option has
+been changed from 30 days to 24 hours.
+
+Certain deleted keys ranges in RocksDB may only get compacted by periodic
+background compactions. The changed default ensures this cleanup to happen more
+frequently. Compactions can potentially lead to spikes in CPU, memory, and I/O
+usage. You may now observe this daily instead of monthly.
+
 ## Client tools
 
 ### arangodump