← Back to the front page
- We’ve added support for airgap installations to the automated installation process by setting a new parameter named
- Multi-channel licenses can now be automated to use a non-default channel using the new param
- Swarm install script now accepts one or more
exclude-subnet flags to prevent Docker from creating networks in a subnet.
- Replicated and Swarm install scripts now accept a
no-ce-on-ee flag which prevents installation of Docker CE on enterprise Linux distributions including RHEL, SLES and Oracle Linux.
- The Replicated install script now accepts the
fast-timeouts flag which sets all input timeouts within the init script to 1 second.
- The Replicated, Repliated UI and Replicated Operator containers now use debian:stretch-slim as their base image. This will result in significantly smaller image sizes.
- The Replicated installation script will now install the
container-selinux package on RHEL 7 when installing Docker, if required.
- Application namespaces when using the Swarm scheduler have been shortened from 43 characters to 12 characters for new installations.
- When installing Docker, the install scripts will now use Docker’s stable channel repos rather than edge.
- The updated support bundle (optionally enabled on the Customer page) has been updated to store Docker logs in two files (stdout, stderr) and also increased compatibility with older Docker versions.
- All known CVEs with patches have been patched in the Docker images Replicated installs.
- Custom publish event timeouts in the Replicated native scheduler were being ignored and set to 10 minutes.
- Restoring from a snapshot on a single host airgap install was not restoring application images.
- Counts of global services shown on the cluster page when using the Swarm scheduler were never reporting more than 1 replica.
- The Swarm node master address shown in the Add Node modal was incorrect on Docker 1.13.
- Within README.md in the v2 support bundle,
VERSION.txt has been replaced by
VERSION.json, matching the file present.
- Preflight checks displayed resource requirements in GiB (2^30), not GB (10^9), which did not match how values were parsed within the application yaml.
- Applications were occasionally failing to start after restoring the nodes of a cluster from a snapshot.
- The Replicated on-prem audit log was broken when SELinux was enabled due to improper SELinux labels being applied to related containers.
test_procs were receiving stale config settings with values using recursive templating.