You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@sdap.apache.org by tl...@apache.org on 2021/04/09 01:17:33 UTC

[incubator-sdap-ingester] 14/33: SDAP-272: Support connecting to Solr through Zk in Granule Ingester (#12)

This is an automated email from the ASF dual-hosted git repository.

tloubrieu pushed a commit to branch ascending_latitudes
in repository https://gitbox.apache.org/repos/asf/incubator-sdap-ingester.git

commit f9462ed6d0e92cf71f3b8cdc13287a627e0da062
Author: Eamon Ford <ea...@gmail.com>
AuthorDate: Wed Aug 5 13:08:21 2020 -0700

    SDAP-272: Support connecting to Solr through Zk in Granule Ingester (#12)
---
 granule_ingester/docker/entrypoint.sh              |  1 +
 granule_ingester/granule_ingester/main.py          | 43 ++++++++----
 .../granule_ingester/writers/SolrStore.py          | 81 ++++++++++++++--------
 granule_ingester/requirements.txt                  |  5 +-
 4 files changed, 85 insertions(+), 45 deletions(-)

diff --git a/granule_ingester/docker/entrypoint.sh b/granule_ingester/docker/entrypoint.sh
index 04ed15c..662bd3d 100644
--- a/granule_ingester/docker/entrypoint.sh
+++ b/granule_ingester/docker/entrypoint.sh
@@ -10,4 +10,5 @@ python /sdap/granule_ingester/main.py \
   $([[ ! -z "$CASSANDRA_USERNAME" ]] && echo --cassandra-username=$CASSANDRA_USERNAME) \
   $([[ ! -z "$CASSANDRA_PASSWORD" ]] && echo --cassandra-password=$CASSANDRA_PASSWORD) \
   $([[ ! -z "$SOLR_HOST_AND_PORT" ]] && echo --solr-host-and-port=$SOLR_HOST_AND_PORT) \
+  $([[ ! -z "$ZK_HOST_AND_PORT" ]] && echo --zk_host_and_port=$ZK_HOST_AND_PORT) \
   $([[ ! -z "$MAX_THREADS" ]] && echo --max-threads=$MAX_THREADS)
diff --git a/granule_ingester/granule_ingester/main.py b/granule_ingester/granule_ingester/main.py
index b54cffd..15390fd 100644
--- a/granule_ingester/granule_ingester/main.py
+++ b/granule_ingester/granule_ingester/main.py
@@ -16,13 +16,14 @@
 import argparse
 import asyncio
 import logging
+import sys
 from functools import partial
 from typing import List
 
 from granule_ingester.consumer import Consumer
+from granule_ingester.exceptions import FailedHealthCheckError, LostConnectionError
 from granule_ingester.healthcheck import HealthCheck
-from granule_ingester.writers import CassandraStore
-from granule_ingester.writers import SolrStore
+from granule_ingester.writers import CassandraStore, SolrStore
 
 
 def cassandra_factory(contact_points, port, username, password):
@@ -31,8 +32,8 @@ def cassandra_factory(contact_points, port, username, password):
     return store
 
 
-def solr_factory(solr_host_and_port):
-    store = SolrStore(solr_host_and_port)
+def solr_factory(solr_host_and_port, zk_host_and_port):
+    store = SolrStore(zk_url=zk_host_and_port) if zk_host_and_port else SolrStore(solr_url=solr_host_and_port)
     store.connect()
     return store
 
@@ -44,7 +45,7 @@ async def run_health_checks(dependencies: List[HealthCheck]):
     return True
 
 
-async def main():
+async def main(loop):
     parser = argparse.ArgumentParser(description='Listen to RabbitMQ for granule ingestion instructions, and process '
                                                  'and ingest a granule for each message that comes through.')
     parser.add_argument('--rabbitmq-host',
@@ -84,6 +85,8 @@ async def main():
                         default='http://localhost:8983',
                         metavar='HOST:PORT',
                         help='Solr host and port. (Default: http://localhost:8983)')
+    parser.add_argument('--zk_host_and_port',
+                        metavar="HOST:PORT")
     parser.add_argument('--max-threads',
                         default=16,
                         metavar='MAX_THREADS',
@@ -111,6 +114,7 @@ async def main():
     cassandra_contact_points = args.cassandra_contact_points
     cassandra_port = args.cassandra_port
     solr_host_and_port = args.solr_host_and_port
+    zk_host_and_port = args.zk_host_and_port
 
     consumer = Consumer(rabbitmq_host=args.rabbitmq_host,
                         rabbitmq_username=args.rabbitmq_username,
@@ -121,19 +125,28 @@ async def main():
                                                    cassandra_port,
                                                    cassandra_username,
                                                    cassandra_password),
-                        metadata_store_factory=partial(solr_factory, solr_host_and_port))
-    if await run_health_checks([CassandraStore(cassandra_contact_points,
-                                               cassandra_port,
-                                               cassandra_username,
-                                               cassandra_password),
-                                SolrStore(solr_host_and_port),
-                                consumer]):
+                        metadata_store_factory=partial(solr_factory, solr_host_and_port, zk_host_and_port))
+    try:
+        solr_store = SolrStore(zk_url=zk_host_and_port) if zk_host_and_port else SolrStore(solr_url=solr_host_and_port)
+        await run_health_checks([CassandraStore(cassandra_contact_points,
+                                                cassandra_port,
+                                                cassandra_username,
+                                                cassandra_password),
+                                 solr_store,
+                                 consumer])
         async with consumer:
             logger.info("All external dependencies have passed the health checks. Now listening to message queue.")
             await consumer.start_consuming(args.max_threads)
-    else:
-        logger.error("Quitting because not all dependencies passed the health checks.")
+    except FailedHealthCheckError as e:
+        logger.error(f"Quitting because not all dependencies passed the health checks: {e}")
+    except LostConnectionError as e:
+        logger.error(f"{e} Any messages that were being processed have been re-queued. Quitting.")
+    except Exception as e:
+        logger.exception(f"Shutting down because of an unrecoverable error:\n{e}")
+    finally:
+        sys.exit(1)
 
 
 if __name__ == '__main__':
-    asyncio.run(main())
+    loop = asyncio.get_event_loop()
+    loop.run_until_complete(main(loop))
diff --git a/granule_ingester/granule_ingester/writers/SolrStore.py b/granule_ingester/granule_ingester/writers/SolrStore.py
index 9d6a7f0..e098672 100644
--- a/granule_ingester/granule_ingester/writers/SolrStore.py
+++ b/granule_ingester/granule_ingester/writers/SolrStore.py
@@ -13,64 +13,87 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-
-from asyncio import AbstractEventLoop
-
+import asyncio
+import functools
+import json
 import logging
+from asyncio import AbstractEventLoop
 from datetime import datetime
 from pathlib import Path
 from typing import Dict
 
-import aiohttp
-from nexusproto.DataTile_pb2 import *
-from tenacity import *
+import pysolr
+from kazoo.handlers.threading import KazooTimeoutError
+from kazoo.exceptions import NoNodeError
+from nexusproto.DataTile_pb2 import TileSummary, NexusTile
 
+from granule_ingester.exceptions import SolrFailedHealthCheckError, SolrLostConnectionError
 from granule_ingester.writers.MetadataStore import MetadataStore
 
 logger = logging.getLogger(__name__)
 
 
+def run_in_executor(f):
+    @functools.wraps(f)
+    def inner(*args, **kwargs):
+        loop = asyncio.get_running_loop()
+        return loop.run_in_executor(None, lambda: f(*args, **kwargs))
+
+    return inner
+
+
 class SolrStore(MetadataStore):
-    def __init__(self, host_and_port='http://localhost:8983'):
+    def __init__(self, solr_url=None, zk_url=None):
         super().__init__()
 
         self.TABLE_NAME = "sea_surface_temp"
         self.iso: str = '%Y-%m-%dT%H:%M:%SZ'
 
-        self._host_and_port = host_and_port
+        self._solr_url = solr_url
+        self._zk_url = zk_url
         self.geo_precision: int = 3
-        self.collection: str = "nexustiles"
+        self._collection: str = "nexustiles"
         self.log: logging.Logger = logging.getLogger(__name__)
         self.log.setLevel(logging.DEBUG)
-        self._session = None
+        self._solr = None
+
+    def _get_connection(self) -> pysolr.Solr:
+        if self._zk_url:
+            zk = pysolr.ZooKeeper(f"{self._zk_url}")
+            collections = {}
+            for c in zk.zk.get_children("collections"):
+                collections.update(json.loads(zk.zk.get("collections/{}/state.json".format(c))[0].decode("ascii")))
+            zk.collections = collections
+            return pysolr.SolrCloud(zk, self._collection, always_commit=True)
+        elif self._solr_url:
+            return pysolr.Solr(f'{self._solr_url}/solr/{self._collection}', always_commit=True)
+        else:
+            raise RuntimeError("You must provide either solr_host or zookeeper_host.")
 
     def connect(self, loop: AbstractEventLoop = None):
-        self._session = aiohttp.ClientSession(loop=loop)
+        self._solr = self._get_connection()
 
     async def health_check(self):
         try:
-            async with aiohttp.ClientSession() as session:
-                response = await session.get('{}/solr/{}/admin/ping'.format(self._host_and_port, self.collection))
-                if response.status == 200:
-                    return True
-                else:
-                    logger.error("Solr health check returned status {}.".format(response.status))
-        except aiohttp.ClientConnectionError as e:
-            logger.error("Cannot connect to Solr!")
-
-        return False
+            connection = self._get_connection()
+            connection.ping()
+        except pysolr.SolrError:
+            raise SolrFailedHealthCheckError("Cannot connect to Solr!")
+        except NoNodeError:
+            raise SolrFailedHealthCheckError("Connected to Zookeeper but cannot connect to Solr!")
+        except KazooTimeoutError:
+            raise SolrFailedHealthCheckError("Cannot connect to Zookeeper!")
 
     async def save_metadata(self, nexus_tile: NexusTile) -> None:
         solr_doc = self._build_solr_doc(nexus_tile)
+        await self._save_document(solr_doc)
 
-        await self._save_document(self.collection, solr_doc)
-
-    @retry(stop=stop_after_attempt(5))
-    async def _save_document(self, collection: str, doc: dict):
-        url = '{}/solr/{}/update/json/docs?commit=true'.format(self._host_and_port, collection)
-        response = await self._session.post(url, json=doc)
-        if response.status < 200 or response.status >= 400:
-            raise RuntimeError("Saving data to Solr failed with HTTP status code {}".format(response.status))
+    @run_in_executor
+    def _save_document(self, doc: dict):
+        try:
+            self._solr.add([doc])
+        except pysolr.SolrError:
+            raise SolrLostConnectionError("Lost connection to Solr, and cannot save tiles.")
 
     def _build_solr_doc(self, tile: NexusTile) -> Dict:
         summary: TileSummary = tile.summary
diff --git a/granule_ingester/requirements.txt b/granule_ingester/requirements.txt
index 4d9d4cb..9b06860 100644
--- a/granule_ingester/requirements.txt
+++ b/granule_ingester/requirements.txt
@@ -1,3 +1,6 @@
 cassandra-driver==3.23.0
-aiomultiprocess
+aiomultiprocess==0.7.0
 aioboto3
+tblib==1.6.0
+pysolr==3.9.0
+kazoo==2.8.0
\ No newline at end of file