From 4f6f8a5e1a02a2aa30d86323f5f9879d7ff433fc Mon Sep 17 00:00:00 2001
From: Daniel Sotirhos <dsotirho@ucsc.edu>
Date: Mon, 6 Jul 2026 14:04:22 -0700
Subject: [PATCH] [H] Fix: Verbatim PFB manifest fails for some datasets in
 anvil14 (#8134)

---
 lambdas/service/.chalice/config.json.template.py |  9 ++++++++-
 src/azul/infra/terraform.py                      | 12 ++++++++++++
 2 files changed, 20 insertions(+), 1 deletion(-)

diff --git a/lambdas/service/.chalice/config.json.template.py b/lambdas/service/.chalice/config.json.template.py
index 374cd7c836..9967ed57ee 100644
--- a/lambdas/service/.chalice/config.json.template.py
+++ b/lambdas/service/.chalice/config.json.template.py
@@ -35,7 +35,14 @@
                 "api_handler": chalice.vpc_lambda_config(app_name),
                 service.generate_manifest.name: {
                     "lambda_timeout": config.service_lambda_timeout,
-                    "lambda_memory_size": 3009 if config.deployment.is_stable else 2048,
+                    # Creating verbatim PFB manifests for large AnVIL datasets
+                    # requires more memory than the default, so we raise it to
+                    # the maximum.
+                    "lambda_memory_size": (
+                        10240 if config.deployment.is_stable and config.is_anvil_enabled()
+                        else 3009 if config.deployment.is_stable
+                        else 2048
+                    ),
                 },
                 service.update_health_cache.name: {
                     "lambda_memory_size": 160,
diff --git a/src/azul/infra/terraform.py b/src/azul/infra/terraform.py
index aa79821d90..82cb2ff51f 100644
--- a/src/azul/infra/terraform.py
+++ b/src/azul/infra/terraform.py
@@ -785,6 +785,18 @@ def resource_items(resource_type: str) -> Iterable[tuple[str, MutableJSON]]:
             package_zip = str(self.package_zip_path(app_name))
             resource['source_code_hash'] = '${filebase64sha256("%s")}' % package_zip
             resource['filename'] = package_zip
+            # Creating verbatim PFB manifests for large AnVIL datasets requires
+            # more ephemeral storage than the default, so we raise it to the
+            # maximum. We have to inject this here since Chalice doesn't support
+            # configuring the ephemeral storage.
+            if (
+                app_name == 'service'
+                and resource_name == f'{app_name}_{config.manifest_sfn}'
+                and config.deployment.is_stable
+                and config.is_anvil_enabled()
+            ):
+                assert 'ephemeral_storage' not in resource
+                resource['ephemeral_storage'] = {'size': 10240}
 
         # Replace any references to unqualified function ARNs in the OpenAPI
         # spec emitted by Chalice with references to the alias.