The release for sprint "Zelda" has been cut. The code for this release is tagged as release-67 in github.
Project Clearwater is in the process of changing the format of data stored in memcached, from a custom binary format to JSON. More detail and background can be found on the Project Clearwater blog: http://www.projectclearwater.org/changing-the-format-of-data-in-memcached. This release is the "checkpoint" release mentioned in the blog post. If you require a seamless upgrade you MUST upgrade to this release. If you do not upgrade, it will not be possible to upgrade to future releases without loss of service (loss of active registrations, and disruption to Rf billing on active calls). In this sprint we have also added code to our C++ components (Sprout, Homestead, Ralf, Memento and Chronos) to protect them against crashes. If a process receives a request that causes it to crash, it will now reject the request (preventing it being automatically retried to other nodes in the cluster). The process will then restart within a maximum of 10 minutes (the actual time of restart is random to protect against all the nodes in a cluster being restarted at the same time). This time is configurable - you can read more about it at https://github.com/Metaswitch/clearwater-docs/wiki/Clearwater-Configuration-Options-Reference. This release also includes the following fixes. * Sprout node quiesce causes a signal 11 crash (https://github.com/Metaswitch/sprout/issues/853) * Clearwater sprout upgrade process needs a way to equalize memcached data to avoid an outage (https://github.com/Metaswitch/sprout/issues/852) * sprout chokes on P-Served-User header with trailing semicolon (https://github.com/Metaswitch/sprout/issues/373) * Configuring an invalid transport on an iFC causes dial-a-boot (https://github.com/Metaswitch/sprout/issues/105) * Impu location lookup request always returns contents of scscf_uri when no HSS is provisioning even when the protocol is not SIP (https://github.com/Metaswitch/homestead/issues/183) * TLS keys can be wiped out (https://github.com/Metaswitch/homestead/issues/73) * Creating keyspaces doesn't wait for Cassandra to start (https://github.com/Metaswitch/homestead/issues/22) * CNAME DNS responses are interpreted as an IPv6 address (https://github.com/Metaswitch/cpp-common/issues/224) * Load monitor options should be configurable (https://github.com/Metaswitch/cpp-common/issues/199) * Memcached store times out connects after 10ms (https://github.com/Metaswitch/cpp-common/issues/38) * chronos.conf.sample includes 'bind_address=0.0.0.0' (https://github.com/Metaswitch/chronos/issues/96) * Chronos uses "auto" type declarations (https://github.com/Metaswitch/chronos/issues/88) * Chronos main.cpp Exit Does Not Appear in ENT Logs (https://github.com/Metaswitch/chronos/issues/87) For upgrading to this release, follow the instructions at https://github.com/Metaswitch/clearwater-docs/wiki/Upgrading-a-Clearwater-Deployment. In this sprint, we've updated Chronos to use our common HTTP Stack. As part of this, Chronos can no longer bind to 0.0.0.0. If you've explicitly set Chronos to do this in the Chronos configuration file (in /etc/chronos/chronos.conf), then you'll need to change this to be the local IP of the Chronos node instead. There are more details about the Chronos configuration file at https://github.com/Metaswitch/chronos/blob/dev/doc/configuration.md. If you are deploying an all-in-one node, the standard image (http://vm-images.cw-ngv.com/cw-aio.ova) has been updated for this release. Thanks, Graeme _______________________________________________ Clearwater mailing list [email protected] http://lists.projectclearwater.org/listinfo/clearwater
