From 950844dc9233f373a365900b36cfe87493dc69ac Mon Sep 17 00:00:00 2001
From: Tessa Walsh <tessa@bitarchivist.net>
Date: Thu, 18 Jan 2024 16:59:40 -0500
Subject: [PATCH] Add migration to fix issues with previous migrations (#1480)

Fixes #1479

- Update null crawlTimeouts in db from null to 0
- Update crawlerChannel in configmaps
---
 backend/btrixcloud/db.py                      |  2 +-
 ...ration_0025_workflow_db_configmap_fixes.py | 49 +++++++++++++++++++
 2 files changed, 50 insertions(+), 1 deletion(-)
 create mode 100644 backend/btrixcloud/migrations/migration_0025_workflow_db_configmap_fixes.py

diff --git a/backend/btrixcloud/db.py b/backend/btrixcloud/db.py
index eb70a8c8..d91c6d2a 100644
--- a/backend/btrixcloud/db.py
+++ b/backend/btrixcloud/db.py
@@ -16,7 +16,7 @@ from pymongo.errors import InvalidName
 from .migrations import BaseMigration
 
 
-CURR_DB_VERSION = "0024"
+CURR_DB_VERSION = "0025"
 
 
 # ============================================================================
diff --git a/backend/btrixcloud/migrations/migration_0025_workflow_db_configmap_fixes.py b/backend/btrixcloud/migrations/migration_0025_workflow_db_configmap_fixes.py
new file mode 100644
index 00000000..6c159e69
--- /dev/null
+++ b/backend/btrixcloud/migrations/migration_0025_workflow_db_configmap_fixes.py
@@ -0,0 +1,49 @@
+"""
+Migration 0025 -- fix workflow database and configmap issues.
+"""
+from btrixcloud.crawlmanager import CrawlManager
+from btrixcloud.migrations import BaseMigration
+from btrixcloud.models import CrawlConfig, UpdateCrawlConfig
+
+
+MIGRATION_VERSION = "0025"
+
+
+class Migration(BaseMigration):
+    """Migration class."""
+
+    def __init__(self, mdb, migration_version=MIGRATION_VERSION):
+        super().__init__(mdb, migration_version)
+
+    async def migrate_up(self):
+        """Perform migration up.
+
+        Set crawlTimeout to 0 in any workflows where it is not set, and
+        update configmap for each workflow to include crawlerChannel.
+        """
+        mdb_crawl_configs = self.mdb["crawl_configs"]
+        try:
+            await mdb_crawl_configs.update_many(
+                {"crawlTimeout": None},
+                {"$set": {"crawlTimeout": 0}},
+            )
+        # pylint: disable=broad-except
+        except Exception:
+            print(
+                "Error updating null crawlconfig crawlTimeouts to 0",
+                flush=True,
+            )
+
+        crawl_manager = CrawlManager()
+        async for config_dict in mdb_crawl_configs.find({}):
+            config = CrawlConfig.from_dict(config_dict)
+            try:
+                await crawl_manager.update_crawl_config(
+                    config, UpdateCrawlConfig(crawlerChannel=config.crawlerChannel)
+                )
+            # pylint: disable=broad-except
+            except Exception as exc:
+                print(
+                    f"Skipping configmap migration for config {config.id} due to error",
+                    exc,
+                )