perf(fulltextsearch): improve elasticsearch healthcheck and add performance tuning settings

- Replace TCP-only nc healthcheck with HTTP cluster health API check via curl,
  so the container is only marked healthy when ES is actually ready (not just
  when the TCP port is open during startup)
- Add indices.fielddata.cache.size=20% to cap field-data cache and prevent
  unbounded heap growth / OOM kills on large datasets
- Add indices.memory.index_buffer_size=20% to improve bulk indexing throughput
  by buffering more data in memory before flushing to disk
- Add thread_pool.write.queue_size=1000 to avoid rejected indexing requests
  when Nextcloud's background jobs submit many files simultaneously

Changes applied to: Containers/fulltextsearch/healthcheck.sh,
php/containers.json, manual-install/latest.yml

Agent-Logs-Url: https://github.com/nextcloud/all-in-one/sessions/a036678d-a9f5-4db3-8cc6-3e3d52f4e97f

Co-authored-by: szaimen <42591237+szaimen@users.noreply.github.com>

Containers/fulltextsearch/healthcheck.sh

@@ -1,3 +1,3 @@
 #!/bin/bash
 
-nc -z 127.0.0.1 9200 || exit 1
+curl -fs "http://127.0.0.1:9200/_cluster/health" | grep -qv '"status":"red"' || exit 1

manual-install/latest.yml

@@ -441,6 +441,9 @@ services:
       - http.port=9200
       - xpack.license.self_generated.type=basic
       - xpack.security.enabled=false
+      - indices.fielddata.cache.size=20%
+      - indices.memory.index_buffer_size=20%
+      - thread_pool.write.queue_size=1000
       - FULLTEXTSEARCH_PASSWORD
     volumes:
       - nextcloud_aio_elasticsearch:/usr/share/elasticsearch/data:rw

php/containers.json

@@ -809,6 +809,9 @@
         "http.port=9200",
         "xpack.license.self_generated.type=basic",
         "xpack.security.enabled=false",
+        "indices.fielddata.cache.size=20%",
+        "indices.memory.index_buffer_size=20%",
+        "thread_pool.write.queue_size=1000",
         "FULLTEXTSEARCH_PASSWORD=%FULLTEXTSEARCH_PASSWORD%"
       ],
       "volumes": [