diff --git a/docs/source/details.rst b/docs/source/details.rst
index d915b428..decd75cf 100644
--- a/docs/source/details.rst
+++ b/docs/source/details.rst
@@ -194,7 +194,7 @@ split data data on the values of those columns. This is done by writing a
 directory structure with *key=value* names. Multiple partition columns can
 be chosen, leading to a multi-level directory tree.
 
-Consider the following directory tree from this `Spark example <http://Spark.apache.org/docs/latest/sql-programming-guide.html#partition-discovery>`_:
+Consider the following directory tree from this `Spark example <https://spark.apache.org/docs/latest/sql-data-sources-parquet.html#partition-discovery>`_:
 
     table/
         gender=male/
@@ -261,4 +261,4 @@ the file system implementation.
 .. raw:: html
 
     <script data-goatcounter="https://fastparquet.goatcounter.com/count"
-        async src="//gc.zgo.at/count.js"></script>
\ No newline at end of file
+        async src="//gc.zgo.at/count.js"></script>