From e3064d101ef8e9074431049135d2319335de3117 Mon Sep 17 00:00:00 2001
From: Stefan Matting <smatting@users.noreply.github.com>
Date: Mon, 15 Feb 2021 18:45:03 +0100
Subject: [PATCH] Add ES migration instructions for release 2021-02-15 (#1364)

---
 .../elasticsearch-migration-2021-02-15.md     | 113 ++++++++++++++++++
 1 file changed, 113 insertions(+)
 create mode 100644 docs/reference/elasticsearch-migration-2021-02-15.md

diff --git a/docs/reference/elasticsearch-migration-2021-02-15.md b/docs/reference/elasticsearch-migration-2021-02-15.md
new file mode 100644
index 00000000000..4f529de83d5
--- /dev/null
+++ b/docs/reference/elasticsearch-migration-2021-02-15.md
@@ -0,0 +1,113 @@
+# ElasticSearch migration instructions for release 2021-02-15
+
+Release `2021-02-15` of `wire-server` requires creating a new ElasticSearch index for `brig` _before_ deploying the release. Without this new index the user search in TeamSettings will be defunct.
+
+The index that brig is using, is defined at brig's config at `elasticsearch.index`. This config value of the previous deployment is referred to as `<OLD_INDEX>` in the following.
+
+These following instructions describe how to:
+
+1. Create the new index `<NEW_INDEX>` (you can choose any name that is different from `<OLD_INDEX>`)
+2. Populate `<NEW_INDEX>` with brig's data.
+3. Configure brig to use the new index.
+
+In the following we assume the following env vars are set.
+
+```bash
+ES_HOST=<YOUR_HOST>
+ES_PORT=<YOUR_PORT> # default is 9200
+OLD_INDEX=<OLD_INDEX> # (Same as `elasticsearch.index` previous deployment)
+NEW_INDEX=<NEW_INDEX> # new index name, different from OLD 
+BRIG_CASSANDRA_HOST=<YOUR_C*_HOST>
+BRIG_CASSANDRA_PORT=<YOUR_C*_PORT>
+BRIG_CASSANDRA_KEYSPACE=<YOUR_C*_KEYSPACE> # probably "brig"
+WIRE_VERSION=<NEW_VERSION_YOU_ARE_DEPLOYING>
+SHARDS=<NUMBER_OF_SHARDS_FOR_THE_INDEX> # 5 if you are unsure
+REPLICAS=<NUMBER_OF_REPLICAS_FOR_THE_INDEX> # 2 if you are unsure
+REFRESH_INTERVAL=<REFRESH_INTERVAL_IN_SECONDS> # 5 if you are unsure
+```
+
+Delete the `directory` index template if it exists:
+
+```bash
+curl -XDELETE http://$ES_HOST:$ES_HOST/_template/directory
+```
+
+The next steps require downtime of brig. If downtime is not acceptable please skip to section [Create the new index without downtime](#create-the-new-index-without-downtime)
+
+1. Shut down the `brig` service.
+2. Create the new index by running
+```sh
+docker run "quay.io/wire/brig-index:$WIRE_VERSION" create \
+    --elasticsearch-server "http://$ES_HOST:$ES_PORT" \
+    --elasticsearch-index "$NEW_INDEX" \
+    --elastcsearch-shards "$SHARDS" \
+    --elastcsearch-replicas "$REPLICAS" \
+    --elastcsearch-refresh-interval "$REFRESH_INTERVAL"
+```
+
+3. Populate the new index by running
+```sh
+docker run "quay.io/wire/brig-index:$WIRE_VERSION" migrate-data \
+  --elasticsearch-server "http://$ES_HOST:$ES_PORT" \
+  --elasticsearch-index "$NEW_INDEX" \
+  --cassandra-host "$BRIG_CASSANDRA_HOST" \
+  --cassandra-port "$BRIG_CASSANDRA_PORT" \
+  --cassandra-keyspace "$BRIG_CASSANDRA_KEYSPACE"
+```
+This may take sevaral minutes depending on number of users.
+
+4. Configure brig to use the new index, by setting `elasticsearch.index` to `<NEW_INDEX>` and deploy of `brig` with new version "$WIRE_VERSION".
+5. Check that team member search in TeamSettings and user search in the app works.
+6. Delete the old index
+```
+```bash
+curl -XDELETE http://$ES_HOST:$ES_HOST/$OLD_INDEX
+```
+
+## Create the new index without downtime
+
+1. Create the new index by running
+
+```sh
+docker run "quay.io/wire/brig-index:$WIRE_VERSION" create \
+    --elasticsearch-server "http://$ES_HOST:$ES_PORT" \
+    --elasticsearch-index "$NEW_INDEX" \
+    --elastcsearch-shards "$SHARDS" \
+    --elastcsearch-replicas "$REPLICAS" \
+    --elastcsearch-refresh-interval "$REFRESH_INTERVAL"
+```
+
+2. Redeploy brig with `elasticsearch.additionalWriteIndex` set to `<NEW_INDEX>`.
+3. Make sure no old instances of brig are running.
+4. Populate the new index by running
+```sh
+docker run "quay.io/wire/brig-index:$WIRE_VERSION" migrate-data \
+  --elasticsearch-server "http://$ES_HOST:$ES_PORT" \
+  --elasticsearch-index "$NEW_INDEX" \
+  --cassandra-host "$BRIG_CASSANDRA_HOST" \
+  --cassandra-port "$BRIG_CASSANDRA_PORT" \
+  --cassandra-keyspace "$BRIG_CASSANDRA_KEYSPACE"
+```
+This may take sevaral minutes depending on number of users.
+
+5. Redeploy brig with these config updates:
+    - unset `elasticsearch.additionalWriteIndex`
+    - set `elasticsearch.index` to `<NEW_INDEX>`
+6. Check that team member search in TeamSettings and user search in the app works.
+7. Delete the old index
+
+```bash
+curl -XDELETE http://$ES_HOST:$ES_HOST/$OLD_INDEX
+```
+
+## Troubleshooting
+
+### Problem: Brig was deployed without creating the new index beforehand.
+Solution: The index can be created after deployment with same instructions as above.
+
+### Problem: `brig-index migrate-data` failed and refuses to migrate again.
+Solution: Delete the migration information from Elasticsearch:
+
+  ```bash
+  curl -XDELETE http://$ES_HOST:$ES_PORT/wire_brig_migrations
+  ```