The release for the sprint "Yojimbo" has now been cut.  This release contains 
the following features and bug fixes:





New features:

* Add support for SIP SUBSCRIBEs and SIP NOTIFYs as per RFC 
3680<http://tools.ietf.org/html/rfc3680>

* Upgrade to Cassandra 1.2





Bug fixes:

* Disabling the local ENUM file also disables ENUM servers 
(https://github.com/Metaswitch/sprout/issues/347)

* Ellis database backup crontab is incorrectly configured 
(https://github.com/Metaswitch/sprout/issues/46)

* SIP-Interface-Specifications page: outdated AKA status 
(https://github.com/Metaswitch/sprout/issues/12)

* Sprout crashes if <Method>REGISTER</Method> is not followed by <Extension> in 
IFCs (https://github.com/Metaswitch/sprout/issues/356)

* Sprout crash if IFCs specify SDP match but no body 
(https://github.com/Metaswitch/sprout/issues/358)

* Log files switch briefly from UTC to local time 
(https://github.com/Metaswitch/sprout/issues/8)

* Memcachedstore is not resilient to memcached server restart 
(https://github.com/Metaswitch/sprout/issues/9)





Upgrade Notes:

The new release of Clearwater ("Yojimbo") includes an upgrade of Apache 
Cassandra, used as the backing database for Homer and Homestead, from version 
1.1 to 1.2. This release of Cassandra includes improvements in clustering 
support, enhancing Clearwater's elastic scaling, as well as several performance 
enhancements. This release also separates out the homestead provisioning 
component from the homestead core component. The homestead provisioning 
component is optional - it is not required if running with an external HSS.



Because of this change, this upgrade includes extra steps: you will have to 
back up your Homer/Homestead database, delete it, upgrade, and recreate it. 
This process is service-affecting, and is expected to take around half an hour 
(based on tests with a system provisioned with 10,000 users). An alternative 
but more complex procedure is available to allow an upgrade without an 
interruption in service, but we're not currently aware of any Clearwater users 
who need uninterrupted service.



Clearwater users who have Homestead configured with an external HSS do not need 
to back the Homestead data up: in this case, subscriber data is mastered in the 
HSS, and the Cassandra database simply stores a local cache. Homestead will 
make a Server-Assignment-Request to retrieve this data when it is next needed.



To perform the upgrade:



*         Upgrade Homestead and install the new Homestead Provisioning package 
on each homestead node by running "sudo apt-get install homestead" and "sudo 
apt-get install homestead-prov".

*         Download the new backup scripts from 
https://github.com/Metaswitch/crest/tree/dev/src/metaswitch/crest/tools/database-dump,
 and copy them into the current directory on one of your Homestead and one of 
your Homer nodes. The backup method documented at 
https://github.com/Metaswitch/clearwater-docs/wiki/Backups does not support 
restoring a Cassanda 1.1 backup onto a Cassandra 1.2 system.

*         If you are not using an external HSS, back up the database on one of 
your homestead nodes using "/usr/share/clearwater/homestead/env/bin/python 
dump-cassandra-to-csv.py homestead_cache homestead_provisioning". This backup 
is processor-intensive and will impact 
performance.https://github.com/Metaswitch/crest/tree/dev/src/metaswitch/crest/tools/database-dump
 gives examples of the expected output.

*         Back up the database on one of your homer nodes using 
"/usr/share/clearwater/homer/env/bin/python dump-cassandra-to-csv.py homer".

*         Run "sudo monit stop cassandra" on every Homer and Homestead node to 
stop Cassandra. This will stop service on your Clearwater deployment until the 
subscriber data is restored at the end of this process.

*         Run 'sudo bash -c "rm -rf /var/lib/cassandra/data*"' on every Homer 
and Homestead node to remove the existing Cassandra database files (which are 
not compatible with 1.2).

*         If you are managing your deployment with Chef, run "git fetch && git 
checkout release-40 && knife cookbook upload clearwater" on the Chef client to 
update the orchestration, then run "sudo chef-client" on all your nodes to 
update the configuration files.

*         Run "sudo clearwater-upgrade" on all nodes - this should restart 
Cassandra and update the Homer/Homestead schemas as an effect of the upgrade. 
When prompted to keep the existing Cassandra files or replace them:

*         If you are managing your deployment with Chef, choose to keep them 
(the default). The previous step has already updated them to 1.2 and filled in 
the correct cluster settings.

*         If you followed the manual install process, choose to replace them.  
This will set up a single-node Cassandra cluster on each Homer and Homestead 
node. If you have multiple Homer nodes or multiple Homestead nodes, and 
previously followed 
https://github.com/Metaswitch/clearwater-docs/wiki/Manual%20Install#clustering-homestead-and-homer
 to cluster them, you will need to repeat that process.

*         Restore from the backups using 
"/usr/share/clearwater/{homer,homestead}/env/bin/python 
restore-cassandra-from-csv.py *.csv.gz" on the Homestead and Homer nodes you 
took the backup on.
*         Finally, tell Ellis to contact the Homestead Provisioning service on 
port 8889 (as opposed to Homestead on port 8888), and open up this port for 
Ellis.
o    If you have a manual install, on Ellis, edit /etc/clearwater/config to add 
a new line "hs_provisioning_hostname=hs.<zone>:8889" (where <zone> is the 
suffix for all of your Clearwater node hostnames) - put another way, 
hs_provisioning_hostname must match hs_hostname but with ":8888" replaced with 
":8889". Then, if you have security groups or firewalls configured between 
Ellis and Homestead, allow TCP traffic from Ellis to homestead on port 8889 
instead of port 8888.
o    If you have done an automated install using chef, Ellis will have been 
configured to use port 8889 when you updated its configuration files earlier. 
Run "knife security groups create -E <name>" to update the security groups.


Your service should now be restored and your systems upgraded to the Yojimbo 
release of Clearwater.

Graeme
_______________________________________________
Clearwater mailing list
[email protected]
http://lists.projectclearwater.org/listinfo/clearwater

Reply via email to