Fixes
- Release limiter slots when a backup is cancelled: previously, cancelled backups kept their limiter slots occupied, preventing subsequent backups from starting by @korotkov-aerospike in #409
- Change the log level for record count errors from ERROR to WARN, as they are not critical in #429
- Remove the internal client ID from log messages to reduce noise in #418
- Fix retry logic for GCP Cloud Storage uploads: retries of large (resumable) uploads were effectively disabled because the retry backoff exceeded the per-chunk retry deadline; transient errors (429/503) during high-parallelism backups are now retried correctly in #476
Improvements
- Add a retry policy for secondary index and UDF creation commands during restore, so transient cluster errors no longer fail the restore in #414
- Implement a scan throttling mechanism to limit the load a backup puts on the cluster in #406
- Reduce memory allocations when encoding backup tokens by @korotkov-aerospike in #431
- Improve the error message shown for encryption configuration issues in #440
- Optimize memory usage for backups to S3 by @korotkov-aerospike in #428
- Internal code cleanup of storage readers, no functional changes by @korotkov-aerospike in #437
- Optimize the restore decoder for faster restore processing by @korotkov-aerospike in #442
- Add validation for the backup file limit to reject invalid values in #448
- Extract estimates printing into a shared library component and print the number of files before starting a restore in #452 and #454
- Show progress with two decimal places (e.g.,
15.67%) and report it more frequently during long backups, so the process doesn't look stalled in #466
Full Changelog: v0.10.0...v0.11.0
Contributors
korotkov-aerospike