Improve init indexing

Tekst-API/.env.template

@@ -73,7 +73,7 @@
 # TEKST_ES__PREFIX=tekst
 # default: tekst
 
-# TEKST_ES__INIT_TIMEOUT_S=120
+# TEKST_ES__INIT_TIMEOUT_S=240
 # default: 120
 
 # TEKST_ES__MAX_FIELD_MAPPINGS=1000

Tekst-API/tekst/config.py

@@ -142,7 +142,7 @@ class ElasticsearchConfig(ConfigSubSection):
     host: str = "127.0.0.1"
     port: int = 9200
     prefix: str = "tekst"
-    init_timeout_s: int = 120
+    init_timeout_s: int = 240
     max_field_mappings: int = 1000
 
     @field_validator("host", mode="before")

Tekst-API/tekst/resources/text_annotation.py

@@ -134,7 +134,7 @@ async def contents_changed_hook(
         cls,
         resource_id: PydanticObjectId,
     ) -> None:
-        op_id = log_op_start(f"Update aggregations for resource {resource_id}")
+        op_id = log_op_start(f"Generate aggregations for resource {resource_id}")
         try:
             await cls._update_aggregations(resource_id)
         except Exception as e:

Tekst-API/tekst/search/__init__.py

@@ -46,30 +46,37 @@
 _es_client: Elasticsearch | None = None
 
 
-async def init_es_client(
-    es_uri: str = _cfg.es.uri,
-) -> Elasticsearch:
+async def _wait_for_es() -> bool:
     global _es_client
-    if _es_client is None:
-        log.info("Initializing Elasticsearch client...")
-        _es_client = Elasticsearch(es_uri)
+    if _es_client is not None:
         for i in range(_cfg.es.init_timeout_s):
             if _es_client.ping():
-                break
+                return True
             if i % 10 == 0:
                 log.debug(
-                    f"Waiting for Elasticsearch service at {es_uri} "
+                    f"Waiting for Elasticsearch service at {_cfg.es.uri} "
                     f"({i}/{_cfg.es.init_timeout_s} seconds)..."
                 )
-            await asyncio.sleep(1)
+                await asyncio.sleep(1)
         else:
-            log.critical(f"Could not connect to Elasticsearch at {es_uri}!")
-            raise RuntimeError("Timed out waiting for Elasticsearch service!")
+            log.critical(f"Could not connect to Elasticsearch at {_cfg.es.uri}!")
+            return False
+    else:
+        await init_es_client()
+
+
+async def init_es_client() -> Elasticsearch:
+    global _es_client
+    if _es_client is None:
+        log.info("Initializing Elasticsearch client...")
+        _es_client = Elasticsearch(_cfg.es.uri)
+        if not await _wait_for_es():
+            raise RuntimeError("Waiting for Elasticsearch client exceeded timeout!")
     return _es_client
 
 
-async def _get_es_client(es_uri: str = _cfg.es.uri) -> Elasticsearch:
-    return await init_es_client(es_uri)
+async def _get_es_client() -> Elasticsearch:
+    return await init_es_client()
 
 
 def get_es_status() -> dict[str, Any] | None:
@@ -107,6 +114,7 @@ async def create_indices_task(
     overwrite_existing_indices: bool = True,
 ) -> dict[str, float]:
     op_id = log_op_start("Create search indices", level="INFO")
+    await _wait_for_es()
     await _setup_index_templates()
 
     # get existing search indices

docs/content/setup/configuration.md

@@ -43,7 +43,7 @@ Configuration for the connection to the Elasticsearch server
 | `TEKST_ES__HOST` | Elasticsearch host (String – default: `127.0.0.1`) |
 | `TEKST_ES__PORT` | Elasticsearch port (Integer – default: `9200`) |
 | `TEKST_ES__PREFIX` | Elasticsearch prefix (for index, templates, etc.) (String – default: `tekst`) |
-| `TEKST_ES__INIT_TIMEOUT_S` | Timeout for waiting for Elasticsearch service to be available on startup (Integer – default: `120`) |
+| `TEKST_ES__INIT_TIMEOUT_S` | Timeout for waiting for Elasticsearch service to be available on startup (Integer – default: `240`) |
 | `TEKST_ES__MAX_FIELD_MAPPINGS` | Max. number of field mappings per search index – given there is enough memory, this can be increased in case there are e.g. annotation resources with many distinct annotation keys (these are dynamically mapped fields). The admin maintenance UI shows a warning if an index is about to hit this value. Any field mapping surpassing this value will be ignored and won't be searchable. (Integer – default: `1000`) |