From e74a812850fdf057ce413df7ee61617f1ad2c1db Mon Sep 17 00:00:00 2001
From: Heikki Linnakangas <heikki.linnakangas@iki.fi>
Date: Wed, 24 Mar 2021 21:50:35 +0200
Subject: [PATCH] Handle WAL redo errors more gracefully.

Don't panic, but return an error to the caller, all the way up to the
processing node instance that did the GetPage@LSN call.
---
 launch.sh           |  0
 src/page_cache.rs   |  9 ++++++++-
 src/page_service.rs | 25 ++++++++++++++++++++-----
 3 files changed, 28 insertions(+), 6 deletions(-)
 mode change 100644 => 100755 launch.sh

diff --git a/launch.sh b/launch.sh
old mode 100644
new mode 100755
diff --git a/src/page_cache.rs b/src/page_cache.rs
index 7e1b1f50b6..997dff98cd 100644
--- a/src/page_cache.rs
+++ b/src/page_cache.rs
@@ -255,7 +255,14 @@ pub fn get_page_at_lsn(tag: BufferTag, lsn: u64) -> Result<Bytes, Box<dyn Error>
             while entry_content.apply_pending {
                 entry_content = entry_rc.walredo_condvar.wait(entry_content).unwrap();
             }
-            page_img = entry_content.page_image.as_ref().unwrap().clone();
+
+            // We should now have a page image. If we don't, it means that WAL redo
+            // failed to reconstruct it. WAL redo should've logged that error already.
+            page_img = match &entry_content.page_image {
+                Some(p) => p.clone(),
+                None => { return Err("could not apply WAL to reconstruct page image".into()); }
+            };
+
         } else {
             // No base image, and no WAL record. Huh?
             return Err(format!("no page image or WAL record for requested page"))?;
diff --git a/src/page_service.rs b/src/page_service.rs
index f811fb260a..8e62a4cf98 100644
--- a/src/page_service.rs
+++ b/src/page_service.rs
@@ -504,11 +504,26 @@ impl Connection {
                     };
 
                     let inf_lsn = 0xffff_ffff_ffff_eeee;
-                    let msg = BeMessage::ZenithReadResponse(ZenithReadResponse {
-                        ok: true,
-                        n_blocks: 0,
-                        page: page_cache::get_page_at_lsn(buf_tag, inf_lsn).unwrap()
-                    });
+
+                    let msg;
+                    {
+                        let p = page_cache::get_page_at_lsn(buf_tag, inf_lsn);
+                        if p.is_ok() {
+                            msg = ZenithReadResponse {
+                                ok: true,
+                                n_blocks: 0,
+                                page: p.unwrap()
+                            };
+                        } else {
+                            const ZERO_PAGE:[u8; 8192] = [0; 8192];
+                            msg = ZenithReadResponse {
+                                ok: false,
+                                n_blocks: 0,
+                                page: Bytes::from_static(&ZERO_PAGE)
+                            }
+                        }
+                    }
+                    let msg = BeMessage::ZenithReadResponse(msg);
 
                     self.write_message(&msg).await?