Add retries to mergepdf (#77)

* update limits * Add retries to mergepdf
lehigh-university-libraries · Jan 20, 2025 · 1c3a177 · 1c3a177
1 parent 619743b
commit 1c3a177
Show file tree

Hide file tree

Showing 6 changed files with 47 additions and 22 deletions.
diff --git a/ci/k8s/crayfits.yaml b/ci/k8s/crayfits.yaml
@@ -27,14 +27,14 @@ spec:
     spec:
       containers:
         - name: scyllaridae-crayfits
-          image: lehighlts/scyllaridae-fits:main
+          image: lehighlts/scyllaridae-fits:main-14a4905
           imagePullPolicy: IfNotPresent
           resources:
             requests:
               memory: "32Mi"
               cpu: "100m"
             limits:
-              memory: "128Mi"
+              memory: "1Gi"
           ports:
             - containerPort: 8080
               hostPort: 8083

diff --git a/ci/k8s/fits.yaml b/ci/k8s/fits.yaml
@@ -8,8 +8,7 @@ spec:
     app: fits
   ports:
     - protocol: TCP
-      port: 8084
-      targetPort: 8080
+      port: 8080
 ---
 apiVersion: apps/v1
 kind: Deployment
@@ -27,7 +26,7 @@ spec:
     spec:
       containers:
         - name: fits
-          image: islandora/fits:3.4
+          image: islandora/fits:3.4.12
           imagePullPolicy: IfNotPresent
           env:
             - name: FITS_MAX_REQUEST_SIZE
@@ -39,10 +38,9 @@ spec:
               memory: "256Mi"
               cpu: "250m"
             limits:
-              memory: "2Gi"
+              memory: "4Gi"
           ports:
             - containerPort: 8080
-              hostPort: 8084
           readinessProbe:
             httpGet:
               path: /fits/version

diff --git a/ci/k8s/houdini.yaml b/ci/k8s/houdini.yaml
@@ -8,7 +8,8 @@ spec:
     app: islandora-imagemagick
   ports:
     - protocol: TCP
-      port: 8080
+      port: 8090
+      targetPort: 8080
 ---
 apiVersion: apps/v1
 kind: Deployment
@@ -26,16 +27,17 @@ spec:
     spec:
       containers:
         - name: scyllaridae-imagemagick
-          image: lehighlts/scyllaridae-imagemagick:main-dff1880
+          image: lehighlts/scyllaridae-imagemagick:main-8a5b743
           imagePullPolicy: IfNotPresent
           resources:
             requests:
-              memory: "128Mi"
+              memory: "256Mi"
               cpu: "100m"
             limits:
-              memory: "2Gi"
+              memory: "10Gi"
           ports:
             - containerPort: 8080
+              hostPort: 8090
           readinessProbe:
             httpGet:
               path: /healthcheck

diff --git a/ci/k8s/mergepdf.yaml b/ci/k8s/mergepdf.yaml
@@ -27,7 +27,7 @@ spec:
     spec:
       containers:
         - name: scyllaridae-mergepdf
-          image: lehighlts/scyllaridae-mergepdf:main-14b2276
+          image: lehighlts/scyllaridae-mergepdf:main-4a2d8ca
           imagePullPolicy: IfNotPresent
           env:
             - name: MAX_THREADS
@@ -37,7 +37,7 @@ spec:
               memory: "128Mi"
               cpu: "500m"
             limits:
-              memory: "4Gi"
+              memory: "24Gi"
           ports:
             - containerPort: 8080
               hostPort: 8088

diff --git a/ci/k8s/whisper.yaml b/ci/k8s/whisper.yaml
@@ -26,14 +26,14 @@ spec:
     spec:
       containers:
         - name: scyllaridae-whisper
-          image: lehighlts/scyllaridae-whisper:main
+          image: lehighlts/scyllaridae-whisper:main-14a4905
           imagePullPolicy: IfNotPresent
           resources:
             requests:
               memory: "2Gi"
               cpu: "500m"
             limits:
-              memory: "5Gi"
+              memory: "10Gi"
               nvidia.com/gpu: "4"
           ports:
             - containerPort: 8080

diff --git a/examples/mergepdf/cmd.sh b/examples/mergepdf/cmd.sh
@@ -6,9 +6,29 @@ TMP_DIR=$(mktemp -d)
 I=0
 MAX_THREADS=${MAX_THREADS:-5}
 PIDS=()
+RETRIES=3
 
-# iterate over all images in the IIIF manifest
-URLS=$(curl -s "$1/book-manifest" | jq -r '.sequences[0].canvases[].images[0].resource."@id"' | awk -F '/' '{print $7}'|sed -e 's/%2F/\//g' -e 's/%3A/:/g')
+# Function to download and process the image with retries
+download_and_process() {
+  local url="$1"
+  local output_file="$2"
+  local attempt=0
+
+  while (( attempt < RETRIES )); do
+    if curl -s "$url" | magick - -resize 1000x\> "$output_file" > /dev/null 2>&1; then
+      return 0
+    fi
+    attempt=$(( attempt + 1 ))
+    echo "Retrying ($attempt/$RETRIES) for $url..."
+    sleep 1
+  done
+
+  echo "Failed to process $url after $RETRIES attempts." >&2
+  return 1
+}
+
+# Iterate over all images in the IIIF manifest
+URLS=$(curl -s "$1/book-manifest" | jq -r '.sequences[0].canvases[].images[0].resource."@id"' | awk -F '/' '{print $7}' | sed -e 's/%2F/\//g' -e 's/%3A/:/g')
 while read -r URL; do
   # If we have reached the max thread limit, wait for any one job to finish
   if [ "${#PIDS[@]}" -ge "$MAX_THREADS" ]; then
@@ -24,11 +44,16 @@ while read -r URL; do
 
   # Run each job in the background
   (
-    # download and resize image to max 1000px width
-    curl -s "$URL" | magick -[0] -resize 1000x\> "$TMP_DIR/img_$I" || curl -s "$URL" | magick - -resize 1000x\> "$TMP_DIR/img_$I" > /dev/null 2>&1
-    # make an OCR'd PDF from the image
-    tesseract "$TMP_DIR/img_$I" "$TMP_DIR/img_$I" pdf > /dev/null 2>&1
-    rm "$TMP_DIR/img_$I"
+    local_img="$TMP_DIR/img_$I"
+
+    # Download and resize the image with retry logic
+    if ! download_and_process "$URL" "$local_img"; then
+      exit 1
+    fi
+
+    # Make an OCR'd PDF from the image
+    tesseract "$local_img" "$TMP_DIR/img_$I" pdf > /dev/null 2>&1
+    rm "$local_img"
   ) &
   PIDS+=("$!")
   I="$(( I + 1))"