diff --git a/BUILD.md b/BUILD.md
index f6f082e..cea63b1 100644
--- a/BUILD.md
+++ b/BUILD.md
@@ -87,10 +87,10 @@ the whole run.
 - **`pyodbc`** is intentionally not built or listed in any target's manifest: modern EL provides
   `python3-pyodbc` from appstream/EPEL, so xcat-dep no longer ships its own. The legacy `pyodbc/`
   directory (an old `pyodbc-3.0.7` RPM spec) is kept for historical reference only.
-- **`conserver`** was replaced by `goconserver` but is provided for completeness and backward
-  compatibility. Core packages depend on `goconserver`; to use `conserver` you must install it
-  explicitly (it is **not** pulled in as a dependency), disable the `goconserver` service and enable
-  the `conserver` service.
+- **`conserver-xcat`** was replaced by `goconserver` but is provided for completeness and backward
+  compatibility. Core packages depend on `goconserver`; to use conserver you must install
+  `conserver-xcat` explicitly (it is **not** pulled in as a dependency), disable the `goconserver`
+  service and enable the `conserver` service.
 
 # Skip and Control Flags
 
diff --git a/mockbuild-all.pl b/mockbuild-all.pl
index 7d9b958..5956a98 100755
--- a/mockbuild-all.pl
+++ b/mockbuild-all.pl
@@ -564,8 +564,8 @@ if (!$skip_build) {
         # or failFast-interrupted prior run leaves the chroot stunted (missing /bin/sh), and mock REUSES
         # the corpse on the next run -> "FileNotFoundError: '/bin/sh'". Scrub it FIRST (best-effort: a
         # no-op on the first run before the config exists) so mock recreates the chroot from the cached
-        # root; --scrub=bootstrap goes too (the per-uniqueext bootstrap is the leak), so the fresh chroot
-        # re-bootstraps from the root cache. mock takes the buildroot lock for --scrub and BLOCKS (not
+        # root; --scrub=bootstrap goes too (the genesis bootstrap chroot is part of the leak; genesis
+        # carries no --mock-uniqueext), so the fresh chroot re-bootstraps from the root cache. mock takes the buildroot lock for --scrub and BLOCKS (not
         # skips) if a concurrent build holds it, but within a run the scrub is sequential before the
         # build and the CD topology never runs a second same-target build at once; `timeout` bounds even
         # a pathological wait so a stale lock can never hang the build.