Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

Excerpt

Impacts for 9.0

  • ES Migration Check - Elasticsearch 2.x has many breaking changes. In particular, if a custom metadata field name includes a period/dot (x-sample-meta-bad.field.name), your existing data cannot be upgraded. Elasticsearch recommends running the migration plugin to check for problems:

    1. Install the migration checker: https://github.com/elastic/elasticsearch-migration/tree/1.x

    2. Browse to http://{Elasticsearch-domain-endpoint}/_plugin/migration/

    3. Select Run checks now.

    4. If problems are found, work with DataCore Support to fix your data.

  • Deprecated and Removed Settings - Remove these settings from your configuration files and use any new settings specified. See Replaced Settings.

    • cluster.proxyPort

    • cluster.proxyIPAddress

    • cluster.settingsUuid

    • disk.automaticFormat

    • security.noauth

  • Deprecated: The native Swarm auth/auth feature is deprecated and is removed as of June 2017. If you are using Swarm's native auth/auth for your applications, you must add security.noauth = False now to continue using the native auth/auth.

  • During a rolling upgrade, you may see SCSP errors on the target cluster ("Castor-System-Cluster value must refer to a remote cluster on RETRIEVE request"); these are harmless and stop when the cluster upgrade is complete. (SWAR-7038)

  • Upgrading from 8.0 replaces several EC and replication settings with the new policy settings. Should you boot an 8.0 node before completing a rolling upgrade to or downgrade from 8.1 or 8.2 (and so have mixed nodes), the new policy settings are removed from the persisted settings and need to be restored using SNMP. See Persisted Settings (SNMP). (SWAR-6901)

...

  • OSS Updates: The Linux kernel is updated to 4.4.14. Firmware updates now include bnx2x (7.12.30.0) and hpsa (3.4.14-0). (SWAR-7049)

  • Improved: The Storage Management UI includes localization into French and Spanish. The popover and chart labels follow your default browser language setting, but they are translated on refresh. (UIS-418)

  • Improved: Contexts (domains and buckets) can no longer be deleted by lifepoint directives, and any delete lifepoints written on these objects are ignored. Use an SCSP recursive delete to remove domains and buckets. (SWAR-6626)

  • Improved: The procedure for updating the administrator password has been streamlined. See Defining Swarm Admins, Swarm Users, and Swarm Passwords. (SWAR-7058)

  • Improved: Swarm processes the backlogs for multiple feeds more equitably. (SWAR-7175)

  • Improved: Swarm log messages include the chassis ID, making it possible to determine which nodes reside on the same chassis when examining logs. (SWAR-6999)

  • Improved: Parallel writes now complete faster, and the response uses chunked transfer encoding to guarantee the client socket remains open in the event of a lengthy complete. The initial response may now be a 202 (Accepted), rather than 201 (Created). For a 202, the completion status is returned in both the body and in the trailing header 'castor-system-result.' For an error, the error message is returned in both the body and the trailing header 'castor-system-error.' (SWAR-7073)

  • Fixed: Any remote replication feed defined in Swarm 8.0 did not propagate existing historical versions of versioned objects to the remote target. (SWAR-6840)

  • Fixed: On cluster reboot, retired and offline drives did not appear in the Swarm Admin Console. (SWAR-7129)

  • Removed: The setting disk.automaticFormat is no longer available. (SWAR-6498)

  • Fixed an issue that prevented volume recovery suspension in the web console. (9.0.1: SWAR-7284)

  • Fixed an issue where FVR/ECR was not kicked off properly in all circumstances. (9.0.1: SWAR-7287)

  • Fixed an issue where nearly full clusters overfill some volumes, leading to 503 write responses. (9.0.2: SWAR-7321)