(tomcat) branch 11.0.x updated: Add a re-try to mitigate against failures observed under load in testing

This is an automated email from the ASF dual-hosted git repository.

markt-asf pushed a commit to branch 11.0.x
in repository https://gitbox.apache.org/repos/asf/tomcat.git


The following commit(s) were added to refs/heads/11.0.x by this push:
     new 4b5656c444 Add a re-try to mitigate against failures observed under load in testing
4b5656c444 is described below

commit 4b5656c444681f6f99b973a6076b8350fc13b47e
Author: Mark Thomas <[email protected]>
AuthorDate: Wed Aug 12 11:20:04 2026 +0100

    Add a re-try to mitigate against failures observed under load in testing
---
 java/org/apache/catalina/session/FileStore.java | 35 +++++++++++++++++++++----
 1 file changed, 30 insertions(+), 5 deletions(-)

diff --git a/java/org/apache/catalina/session/FileStore.java b/java/org/apache/catalina/session/FileStore.java
index 6ca0d521c5..4df5bb1faa 100644
--- a/java/org/apache/catalina/session/FileStore.java
+++ b/java/org/apache/catalina/session/FileStore.java
@@ -37,6 +37,7 @@ import org.apache.catalina.Context;
 import org.apache.catalina.Session;
 import org.apache.juli.logging.Log;
 import org.apache.juli.logging.LogFactory;
+import org.apache.tomcat.util.ExceptionUtils;
 import org.apache.tomcat.util.concurrent.KeyedReentrantReadWriteLock;
 import org.apache.tomcat.util.res.StringManager;
 
@@ -281,11 +282,35 @@ public final class FileStore extends StoreBase {
                     ObjectOutputStream oos = new ObjectOutputStream(new BufferedOutputStream(fos))) {
                 ((StandardSession) session).writeObjectData(oos);
             }
-            try {
-                Files.move(tempFile.toPath(), file.toPath(), StandardCopyOption.REPLACE_EXISTING,
-                        StandardCopyOption.ATOMIC_MOVE);
-            } catch (AtomicMoveNotSupportedException e) {
-                Files.move(tempFile.toPath(), file.toPath(), StandardCopyOption.REPLACE_EXISTING);
+            /*
+             * Failures have been observed with the move when under load in testing. The re-try mechanism is an attempt
+             * to mitigate against those failures.
+             */
+            int attempts = 0;
+            int maxAttempts = 2;
+            while (attempts < maxAttempts) {
+                attempts++;
+                try {
+                    try {
+                        Files.move(tempFile.toPath(), file.toPath(), StandardCopyOption.REPLACE_EXISTING,
+                                StandardCopyOption.ATOMIC_MOVE);
+                    } catch (AtomicMoveNotSupportedException e) {
+                        Files.move(tempFile.toPath(), file.toPath(), StandardCopyOption.REPLACE_EXISTING);
+                    }
+                    break;
+                } catch (Throwable t) {
+                    ExceptionUtils.handleThrowable(t);
+                    if (attempts < maxAttempts) {
+                        // Brief delay before re-try
+                        try {
+                            Thread.sleep(50);
+                        } catch (InterruptedException e) {
+                            // Ignore. The delay will just be shorter than expected.
+                        }
+                    } else {
+                        throw t;
+                    }
+                }
             }
         } finally {
             try {