This release focuses on improvements to support Amazon S3 compatibility and to match Amazon S3 behavior:
Swarm Storage 9.3+ Required — Content Gateway 5.2.3 supports Storage 9.3 and newer. The storage cluster must be upgraded prior to installing this version of Gateway. If you are using Storage 9.4, update to Storage UI 1.2.3, which corrects an issue with defining a replication feed that specifies domains. (CLOUD-2890)
Gateway 5.2.2 requires Swarm Storage 9.3+.
Gateway 5.2.0 requires Swarm Storage 9.2.1 and is not compatible with Swarm Storage 9.3.
Gateway 5.1.3 and earlier are not compatible with Swarm Storage 9.2 and higher. For additional historic compatibility details, refer to the releases included with those distributions.
EC required — Erasure coding (EC) must be enabled and configured; if not, the Gateway cannot start (
service cloudgateway init fails). (CLOUD-2587)
Elasticsearch 2.3.3 — Swarm 9 includes significant enhancements to its use of Elasticsearch and Gateway now depends upon features from Elasticsearch 2.3. Prior to upgrading Gateway, you must complete the migration to Elasticsearch 2.3.
RPM changes — The Content Gateway RPM package has changed from "caringo-cloudgateway" to "caringo-gateway". (CLOUD-2728)
Re-enable service — After an upgrade from version 5.2.1 or earlier, the service does not automatically start after a system reboot. To re-enable the service, run "chkconfig --add cloudgateway" (RHEL 6) or "systemctl enable cloudgateway" (RHEL 7). (5.2.2: CLOUD-2819)
ExpanDrive — ExpanDrive users must upgrade to ExpanDrive 6.1.0 and Gateway 5.2.2; earlier versions report 403 Signature errors when creating a folder or uploading a large object. (5.2.2: CLOUD-2746)
Improved: A new [gateway] setting, cookieDomains, allows the Content Portal to use the same authentication token across multiple storage domains that share a common base domain. Gateway does this by using the base domain in place of the request's domain for the Set-Cookie response header. (5.2.2: CLOUD-2789)
Improved: Gateway has better handling for requests to domains that are marked as belonging to an unknown tenant. These requests are handled as if they are from the "System Tenant" and troubleshooting guidance is recorded in the application logs. (5.2.1: CLOUD-2714)
Improved: Gateway has lowered the default request rewind buffer to 64KB to prevent an excessive volume of messages in the storage cluster logs and to match the storage cluster's upper threshold after which 100-continue must be used. This should have no impact for applications using Gateway. (CLOUD-2652)
Fixed: A versioned listing now responds in the correct format when the bucket is not versioned, fixing a problem that occurred with Cyberduck 6.2.3 and later. (5.2.2: CLOUD-2796)
Fixed: When using a "pam" idsys for Linux users on the Gateway server, requests may slow down after days of high, sustained load. (5.2.1: CLOUD-2727)
Fixed: S3 V4 signature errors occurred when using X-Forwarded-Host on a request. Setting [debug] setForwarded=false prevents these errors, if they occur. (5.2.1: CLOUD-2706)
Fixed: Multipart/form-data MIME POST uploads that spanned several days of sustained load can result in an OutOfMemoryError. (5.2.1: CLOUD-2732)
Fixed: Gateway startup can silently hang if there is a problem communicating with the indexerHosts or if Swarm is not configured with a Search feed. (5.2.1: CLOUD-2419)
Fixed: Gateway log warning messages about "Swarm connection failure to node 127.0.0.1" can be generated when using Swarm UI. (5.2.1: CLOUD-2749)
Fixed: Single-shot context creation failed in Gateway 5.1.2 and 5.1.3; the POST failed with a 500 "JSON Validation Errors". (CLOUD-2703)
Fixed: In the Content Portal, a collection can appear as a bucket, such that deleting the collection can delete the bucket it was scoped to. For collections erroneously showing the bucket icon, make a small edit and resave them, which causes them to be fully repaired. (CLOUD-2694, CLOUD-2712)