Migrate to OpenSearch (#16180)

* use opensearch container locally, to match production index

* migrate to opensearchpy

* rename os -> opensearch, no need to save keystrokes

* Rename wrapper/page

* rename env var

* update docs

* remove refs to dead deps

Author: ewdurbin

.github/dependabot.yml

@@ -11,8 +11,6 @@ updates:
   - dependency-type: indirect
   rebase-strategy: "disabled"
   ignore:
-  # Always ignore elasticsearch, future versions are always incompatible with our provider
-  - dependency-name: "elasticsearch"
   # These update basically every day, and 99.9% of the time we don't care
   - dependency-name: "boto3"
   - dependency-name: "boto3-stubs"

dev/compose/opensearch/Dockerfile

@@ -0,0 +1,7 @@
+FROM opensearchproject/opensearch:2.12.0
+
+RUN opensearch-plugin remove opensearch-skills --purge
+RUN opensearch-plugin remove opensearch-ml --purge
+RUN opensearch-plugin remove opensearch-neural-search --purge
+RUN opensearch-plugin remove opensearch-performance-analyzer --purge
+RUN opensearch-plugin remove opensearch-security-analytics --purge

dev/environment

@@ -11,7 +11,7 @@ BROKER_URL=sqs://localstack:4566/?region=us-east-1&queue_name_prefix=warehouse-d
 
 DATABASE_URL=postgresql+psycopg://postgres@db/warehouse
 
-ELASTICSEARCH_URL=http://elasticsearch:9200/development
+OPENSEARCH_URL=http://opensearch:9200/development
 
 REDIS_URL=redis://redis:6379/0
 

docker-compose.override.yaml-sample

@@ -34,7 +34,7 @@ services:
   dev-docs: *disable-service
   user-docs: *disable-service
 
-  elasticsearch:
+  opensearch:
     # You can also add selective environment variables
     environment:
       logger.level: WARN  # default INFO is pretty noisy

docker-compose.yml

@@ -44,11 +44,20 @@ services:
     ports:
       - "4566:4566"
 
-  elasticsearch:
-    image: elasticsearch:7.10.1
+  opensearch:
+    build:
+      context: ./dev/compose/opensearch
+    init: true
+    healthcheck:
+        test: ["CMD-SHELL", "curl -u admin:gqYeDIzbEwTTYmB7 --silent --fail http://localhost:9200/_cluster/health || exit 1"]
+        interval: 1s
+        start_period: 10s
     environment:
-      - xpack.security.enabled=false
       - discovery.type=single-node
+      - OPENSEARCH_INITIAL_ADMIN_PASSWORD=gqYeDIzbEwTTYmB7
+      - DISABLE_INSTALL_DEMO_CONFIG=true
+      - DISABLE_SECURITY_PLUGIN=true
+      - DISABLE_PERFORMANCE_ANALYZER_AGENT_CLI=true
     ulimits:
       nofile:
         soft: 65536
@@ -109,7 +118,7 @@ services:
     depends_on:
       db:
         condition: service_healthy
-      elasticsearch:
+      opensearch:
         condition: service_started
       redis:
         condition: service_started

docs/dev/architecture.md

@@ -91,12 +91,12 @@ C4Container
     Container(camo, "Camo", "image proxy")
     Container(web_app, "Web", "Python (Pyramid, SQLAlchemy)", "Delivers HTML and API content")
     SystemQueue(sqs, "AWS SQS", "task broker")
-    SystemDb(elasticsearch, "Elasticsearch", "Index of projects, packages, metadata")
+    SystemDb(opensearch, "OpenSearch", "Index of projects, packages, metadata")
     SystemDb(db, "Postgres Database", "Store project, package metadata, user details")
     SystemDb(redis, "Redis", "Store short-term cache data")
 
     Rel(web_app, sqs, "queue tasks")
-    Rel(web_app, elasticsearch, "search for projects")
+    Rel(web_app, opensearch, "search for projects")
     Rel(web_app, db, "store/retrieve most data")
     Rel(web_app, redis, "cache data")
   }
@@ -153,7 +153,7 @@ C4Container
   Container_Boundary(c1, "Supporting Systems") {
     SystemDb(redis, "Redis", "Store short-term cache data")
     SystemQueue(sqs, "AWS SQS", "task broker")
-    SystemDb(elasticsearch, "Elasticsearch", "Index of projects, packages, metadata")
+    SystemDb(opensearch, "OpenSearch", "Index of projects, packages, metadata")
     SystemDb(db, "Postgres Database", "Store project, package metadata, user details")
     System(ses, "AWS SES", "Simple Email Service")
   }
@@ -163,7 +163,7 @@ C4Container
   BiRel(worker, sqs, "get next task/ack")
   BiRel(worker, redis, "store task results")
   BiRel(worker, db, "interact with models")
-  BiRel(worker, elasticsearch, "update search index")
+  BiRel(worker, opensearch, "update search index")
   Rel(worker, fastly, "purge URLs")
   Rel(worker, ses, "send emails")
 

docs/dev/development/getting-started.rst

@@ -188,8 +188,8 @@ application.
    (on Windows by editing the config file found at ``C:\Users\<USER>\AppData\Local\Docker\wsl``).
 
    If you are using Linux, you may need to configure the maximum map count to get
-   the `elasticsearch` up and running. According to the
-   `documentation <https://www.elastic.co/guide/en/elasticsearch/reference/6.8/vm-max-map-count.html>`_
+   the `opensearch` up and running. According to the
+   `documentation <https://opensearch.org/docs/2.15/install-and-configure/install-opensearch/index/#important-settings>`_
    this can be set temporarily:
 
    .. code-block:: console
@@ -200,9 +200,9 @@ application.
    :file:`/etc/sysctl.conf`.
 
    Also check that you have more than 5% disk space free, otherwise
-   elasticsearch will become read only. See ``flood_stage`` in the
-   `elasticsearch disk allocation docs
-   <https://www.elastic.co/guide/en/elasticsearch/reference/6.8/disk-allocator.html>`_.
+   opensearch will become read only. See ``flood_stage`` in the
+   `opensearch disk allocation docs
+   <https://opensearch.org/docs/latest/install-and-configure/configuring-opensearch/cluster-settings/#cluster-level-routing-and-allocation-settings>`_.
 
 
 Once ``make build`` has finished,  run the command:
@@ -414,10 +414,10 @@ Errors when executing ``make initdb``
 
 * If ``make initdb`` fails with a timeout like::
 
-    urllib3.exceptions.ConnectTimeoutError: (<urllib3.connection.HTTPConnection object at 0x8beca733c3c8>, 'Connection to elasticsearch timed out. (connect timeout=30)')
+    urllib3.exceptions.ConnectTimeoutError: (<urllib3.connection.HTTPConnection object at 0x8beca733c3c8>, 'Connection to opensearch timed out. (connect timeout=30)')
 
   you might need to increase the amount of memory allocated to docker, since
-  elasticsearch wants a lot of memory (Dustin gives warehouse ~4GB locally).
+  opensearch wants a lot of memory (Dustin gives warehouse ~4GB locally).
   Refer to the tip under :ref:`running-warehouse-containers` section for more details.
 
 
@@ -478,7 +478,7 @@ Docker please raise an issue in
 Disabling services locally
 ^^^^^^^^^^^^^^^^^^^^^^^^^^
 
-Some services, such as Elasticsearch, consume a lot of resources when running
+Some services, such as OpenSearch, consume a lot of resources when running
 locally, but might not always be necessary when doing local development.
 
 To disable these locally, you can create a ``docker-compose.override.yaml``
@@ -490,8 +490,8 @@ individually disable services, modify their entrypoint to do something else:
     version: "3"
 
     services:
-      elasticsearch:
-        entrypoint: ["echo", "Elasticsearch disabled"]
+      opensearch:
+        entrypoint: ["echo", "OpenSearch disabled"]
 
 Note that disabling services might cause things to fail in unexpected ways.
 

pyproject.toml

@@ -56,7 +56,6 @@ module = [
     "b2sdk.*",  # https://github.com/Backblaze/b2-sdk-python/issues/148
     "celery.app.backends.*",
     "celery.backends.redis.*",
-    "elasticsearch_dsl.*", # https://github.com/elastic/elasticsearch-dsl-py/issues/1533
     "github_reserved_names.*",
     "google.cloud.*",
     "forcediphttpsadapter.*",

requirements/main.in

@@ -13,8 +13,6 @@ click
 cryptography
 datadog>=0.19.0
 disposable-email-domains
-elasticsearch>=7.0.0,<7.11.0
-elasticsearch_dsl>=7.0.0,<8.0.0
 first
 forcediphttpsadapter
 github-reserved-names>=1.0.0
@@ -31,6 +29,7 @@ linehaul
 lxml
 msgpack
 natsort
+opensearch-py
 orjson
 packaging>=23.2
 packaging_legacy
@@ -69,7 +68,7 @@ structlog
 transaction
 trove-classifiers
 ua-parser
-urllib3<2  # See https://github.com/pypi/warehouse/issues/14671
+urllib3
 webauthn>=1.0.0,<3.0.0
 whitenoise
 WTForms[email]>=2.0.0

requirements/main.txt

@@ -184,7 +184,7 @@ certifi==2024.6.2 \
     --hash=sha256:ddc6c8ce995e6987e7faf5e3f1b02b302836a0e5d98ece18392cb1a36c72ad56
     # via
     #   -r requirements/main.in
-    #   elasticsearch
+    #   opensearch-py
     #   requests
     #   sentry-sdk
 cffi==1.16.0 \
@@ -485,20 +485,13 @@ docutils==0.20.1 \
     --hash=sha256:96f387a2c5562db4476f09f13bbab2192e764cac08ebbf3a34a95d9b1e4a59d6 \
     --hash=sha256:f08a4e276c3a1583a86dce3e34aba3fe04d02bba2dd51ed16106244e8a923e3b
     # via readme-renderer
-elasticsearch==7.10.1 \
-    --hash=sha256:4ebd34fd223b31c99d9f3b6b6236d3ac18b3046191a37231e8235b06ae7db955 \
-    --hash=sha256:a725dd923d349ca0652cf95d6ce23d952e2153740cf4ab6daf4a2d804feeed48
-    # via
-    #   -r requirements/main.in
-    #   elasticsearch-dsl
-elasticsearch-dsl==7.4.1 \
-    --hash=sha256:07ee9c87dc28cc3cae2daa19401e1e18a172174ad9e5ca67938f752e3902a1d5 \
-    --hash=sha256:97f79239a252be7c4cce554c29e64695d7ef6a4828372316a5e5ff815e7a7498
-    # via -r requirements/main.in
 email-validator==2.2.0 \
     --hash=sha256:561977c2d73ce3611850a06fa56b414621e0c8faa9d66f2611407d87465da631 \
     --hash=sha256:cb690f344c617a714f22e66ae771445a1ceb46821152df8e165c5f9a364582b7
     # via wtforms
+events==0.5 \
+    --hash=sha256:a7286af378ba3e46640ac9825156c93bdba7502174dd696090fdfcd4d80a1abd
+    # via opensearch-py
 first==2.0.2 \
     --hash=sha256:8d8e46e115ea8ac652c76123c0865e3ff18372aef6f03c22809ceefcea9dec86 \
     --hash=sha256:ff285b08c55f8c97ce4ea7012743af2495c9f1291785f163722bd36f6af6d3bf
@@ -1285,6 +1278,10 @@ openapi-spec-validator==0.7.1 \
     --hash=sha256:3c81825043f24ccbcd2f4b149b11e8231abce5ba84f37065e14ec947d8f4e959 \
     --hash=sha256:8577b85a8268685da6f8aa30990b83b7960d4d1117e901d451b5d572605e5ec7
     # via openapi-core
+opensearch-py==2.6.0 \
+    --hash=sha256:0b7c27e8ed84c03c99558406927b6161f186a72502ca6d0325413d8e5523ba96 \
+    --hash=sha256:b6e78b685dd4e9c016d7a4299cf1de69e299c88322e3f81c716e6e23fe5683c1
+    # via -r requirements/main.in
 orjson==3.10.5 \
     --hash=sha256:03b565c3b93f5d6e001db48b747d31ea3819b89abf041ee10ac6988886d18e01 \
     --hash=sha256:099e81a5975237fda3100f918839af95f42f981447ba8f47adb7b6a3cdb078fa \
@@ -1649,8 +1646,8 @@ python-dateutil==2.9.0.post0 \
     #   botocore
     #   celery
     #   celery-redbeat
-    #   elasticsearch-dsl
     #   google-cloud-bigquery
+    #   opensearch-py
 python-slugify==8.0.4 \
     --hash=sha256:276540b79961052b66b7d116620b36518847f52d5fd9e3a70164fc8c50faa6b8 \
     --hash=sha256:59202371d1d05b54a9e7720c5e038f928f45daaffe41dd10822f3907b937c856
@@ -1745,6 +1742,7 @@ requests==2.32.3 \
     #   google-cloud-bigquery
     #   google-cloud-storage
     #   jsonschema-path
+    #   opensearch-py
     #   premailer
     #   requests-aws4auth
     #   stripe
@@ -1880,9 +1878,9 @@ six==1.16.0 \
     --hash=sha256:8abb2f1d86890a2dfb989f9a77cfcfd3e47c2a354b01111771326f8aa26e0254
     # via
     #   automat
-    #   elasticsearch-dsl
     #   html5lib
     #   isodate
+    #   opensearch-py
     #   pymacaroons
     #   python-dateutil
     #   requests-aws4auth
@@ -2006,8 +2004,8 @@ urllib3==1.26.19 \
     #   -r requirements/main.in
     #   botocore
     #   celery
-    #   elasticsearch
     #   kombu
+    #   opensearch-py
     #   requests
     #   sentry-sdk
 venusian==3.1.0 \

tests/conftest.py

@@ -312,7 +312,7 @@ def app_config(database):
         "database.url": database,
         "docs.url": "http://docs.example.com/",
         "ratelimit.url": "memory://",
-        "elasticsearch.url": "https://localhost/warehouse",
+        "opensearch.url": "https://localhost/warehouse",
         "files.backend": "warehouse.packaging.services.LocalFileStorage",
         "archive_files.backend": "warehouse.packaging.services.LocalArchiveFileStorage",
         "simple.backend": "warehouse.packaging.services.LocalSimpleStorage",

tests/unit/search/test_init.py

@@ -10,7 +10,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import elasticsearch
+import opensearchpy
 import pretend
 
 from warehouse import search
@@ -69,7 +69,7 @@ def test_execute_unindex_success(app_config):
     assert "warehouse.search.project_deletes" not in session.info
 
 
-def test_es(monkeypatch):
+def test_opensearch(monkeypatch):
     search_obj = pretend.stub()
     index_obj = pretend.stub(
         document=pretend.call_recorder(lambda d: None),
@@ -84,15 +84,15 @@ def test_es(monkeypatch):
     client = pretend.stub()
     request = pretend.stub(
         registry={
-            "elasticsearch.client": client,
-            "elasticsearch.index": "warehouse",
+            "opensearch.client": client,
+            "opensearch.index": "warehouse",
             "search.doc_types": doc_types,
         }
     )
 
-    es = search.es(request)
+    opensearch = search.opensearch(request)
 
-    assert es is search_obj
+    assert opensearch is search_obj
     assert index_cls.calls == [pretend.call("warehouse", using=client)]
     assert index_obj.document.calls == [pretend.call(d) for d in doc_types]
     assert index_obj.settings.calls == [
@@ -104,20 +104,20 @@ def test_es(monkeypatch):
 def test_includeme(monkeypatch):
     aws4auth_stub = pretend.stub()
     aws4auth = pretend.call_recorder(lambda *a, **kw: aws4auth_stub)
-    es_client = pretend.stub()
-    es_client_init = pretend.call_recorder(lambda *a, **kw: es_client)
+    opensearch_client = pretend.stub()
+    opensearch_client_init = pretend.call_recorder(lambda *a, **kw: opensearch_client)
 
     monkeypatch.setattr(search.requests_aws4auth, "AWS4Auth", aws4auth)
-    monkeypatch.setattr(search.elasticsearch, "Elasticsearch", es_client_init)
+    monkeypatch.setattr(search.opensearchpy, "OpenSearch", opensearch_client_init)
 
     registry = {}
-    es_url = "https://some.url/some-index?aws_auth=1&region=us-east-2"
+    opensearch_url = "https://some.url/some-index?aws_auth=1&region=us-east-2"
     config = pretend.stub(
         registry=pretend.stub(
             settings={
                 "aws.key_id": "AAAAAAAAAAAA",
                 "aws.secret_key": "deadbeefdeadbeefdeadbeef",
-                "elasticsearch.url": es_url,
+                "opensearch.url": opensearch_url,
             },
             __setitem__=registry.__setitem__,
         ),
@@ -130,20 +130,20 @@ def test_includeme(monkeypatch):
     assert aws4auth.calls == [
         pretend.call("AAAAAAAAAAAA", "deadbeefdeadbeefdeadbeef", "us-east-2", "es")
     ]
-    assert len(es_client_init.calls) == 1
-    assert es_client_init.calls[0].kwargs["hosts"] == ["https://some.url"]
-    assert es_client_init.calls[0].kwargs["timeout"] == 2
-    assert es_client_init.calls[0].kwargs["retry_on_timeout"] is False
+    assert len(opensearch_client_init.calls) == 1
+    assert opensearch_client_init.calls[0].kwargs["hosts"] == ["https://some.url"]
+    assert opensearch_client_init.calls[0].kwargs["timeout"] == 2
+    assert opensearch_client_init.calls[0].kwargs["retry_on_timeout"] is False
     assert (
-        es_client_init.calls[0].kwargs["connection_class"]
-        == elasticsearch.connection.http_requests.RequestsHttpConnection
+        opensearch_client_init.calls[0].kwargs["connection_class"]
+        == opensearchpy.connection.http_requests.RequestsHttpConnection
     )
-    assert es_client_init.calls[0].kwargs["http_auth"] == aws4auth_stub
+    assert opensearch_client_init.calls[0].kwargs["http_auth"] == aws4auth_stub
 
-    assert registry["elasticsearch.client"] == es_client
-    assert registry["elasticsearch.index"] == "some-index"
-    assert registry["elasticsearch.shards"] == 1
-    assert registry["elasticsearch.replicas"] == 0
+    assert registry["opensearch.client"] == opensearch_client
+    assert registry["opensearch.index"] == "some-index"
+    assert registry["opensearch.shards"] == 1
+    assert registry["opensearch.replicas"] == 0
     assert config.add_request_method.calls == [
-        pretend.call(search.es, name="es", reify=True)
+        pretend.call(search.opensearch, name="opensearch", reify=True)
     ]