AWS solutions documentation change
Summary
Updated troubleshooting table: modified existing entry for Solr repository errors, added 10 new common migration issues and resolutions covering replayer configuration, schema handling, proxy setup, and backfill errors. Updated related information links.
Security assessment
Changes are operational troubleshooting steps for migration tooling. No vulnerabilities, security patches, or security weaknesses are addressed. The subnet tagging resolution (kubernetes.io/role/elb) is a standard infrastructure requirement without security implications.
Diff
diff --git a/solutions/latest/migration-assistant-for-amazon-opensearch-service/solr-troubleshooting.md b/solutions/latest/migration-assistant-for-amazon-opensearch-service/solr-troubleshooting.md index 5c02d7f90..2234f9b04 100644 --- a//solutions/latest/migration-assistant-for-amazon-opensearch-service/solr-troubleshooting.md +++ b//solutions/latest/migration-assistant-for-amazon-opensearch-service/solr-troubleshooting.md @@ -12 +12 @@ Symptom | Resolution -`Repository default-s3 not found` (or `Repository s3 not found`) | The `<backup>` block with the S3 `<repository>` is missing from `solr.xml`, or `solr.xml` was not republished after editing. Add the block, publish `solr.xml`, and restart. +`Repository default-s3 not found` (or `Repository s3 not found`) | The `backup` block with the S3 `repository` is missing from `solr.xml`, or `solr.xml` was not republished after editing. Add the block, publish `solr.xml`, and restart. @@ -13,0 +14,10 @@ Symptom | Resolution +Replayer running but `requests=0` | The replayer image does not include Solr transform logic. Verify the replayer image includes the Solr transform JARs. Check the `migration-image-config` ConfigMap. +"Could not find provider: SolrToOpenSearchTransformProvider" | Transform JAR not on replayer classpath. Ensure the replayer image was built with Solr transform overlay. Re-run the deploy pipeline if needed. +"Failed to parse solrconfig.xml: Content is not allowed in prolog" | File contains ZooKeeper warning output or is JSON from the `/config` endpoint. Re-extract using `solr zk cp` to a temp file, then copy the temp file. +Queries for one collection replay with the wrong `match` or `term` behavior | The request is probably using the default schema instead of the collection-specific schema. Add `collection.solrSchemaXml` for that collection, and make sure `collection` exactly matches the name in request paths such as `/solr/collection/select`. +Solr request-handler defaults, invariants, or appends are not reflected in replayed queries | The matching `solrconfig.xml` was not loaded for that collection. Confirm the ConfigMap key and `fromFile.path`, make sure the file is XML copied from ZooKeeper rather than JSON from the `/config` endpoint, and check the replayer logs for `Loaded solrConfig` or parse-warning messages. +`targetResponses={400}` with "refresh policy not supported" | `targetType` not set for a Serverless target. Set `"targetType": {"value": "OpenSearchServerless"}` or `"NextGenOpenSearchServerless"` in the transform provider’s context values. +`No zk_backup directories found for collection` in metadata or backfill logs | The backup is usually staged at the wrong S3 level or is missing the ZooKeeper backup. Set `s3RepoPathUri` to the parent path, set `externallyManagedSnapshotName` to the backup root, and verify that each collection directory contains `zk_backup_N/` or `zk_backup/` as described in [Verify the Solr backup layout](./solr-prepare-backup.html#solr-backup-layout). +Metadata migration creates empty or mostly fallback mappings for a Solr collection | The collection’s latest ZooKeeper backup does not include a readable schema file. Confirm that `managed-schema.xml`, `managed-schema`, or `schema.xml` exists under the collection’s latest `zk_backup_N/configs/CONFIGSET/` directory. +NLB for capture proxy stuck in `pending` | Subnets missing the required ELB tag. Tag subnets: `aws ec2 create-tags --resources SUBNET_IDs --tags Key=kubernetes.io/role/elb,Value=1`. +Kafka entity-operator CrashLoopBackOff | Startup probe timeout during initial Kafka provisioning. Delete the pod to reset backoff: `kubectl delete pod -n ma -l strimzi.io/name=default-entity-operator`. @@ -23 +33 @@ To use the Amazon Web Services Documentation, Javascript must be enabled. Please -Configure and run the backfill workflow +Capture and replay live traffic from Solr @@ -25 +35 @@ Configure and run the backfill workflow -Playbooks +Transform data and requests