Upgrade instructions

This instuctions describe how to upgrade Yeti system from version 1.7.15 to version 1.8. If you have any other Yeti version you should previously upgrade it to 1.7.15

Get know your postgresql cluster epoch

PGQ components of CDR database require cluster epoch increment. To get know your cluster epoch, connect to CDR postgresql database using psql and run:

cdr=# SELECT (txid_current() >> 32) as epoch;
epoch
-------
        0
(1 row)

See https://github.com/markokr/skytools/blob/master/doc/faq.txt#L50 for details

Upgrade yeti-web package to 1.8.0

root@yeti:/# apt update
root@yeti:/# apt install yeti-web=1.8.0

Postgresql 11 installation

You should install postgresql-11 packages, as described at Routing database installation and CDR databases installation and create appropriate databases.

Run new instances of Postgresql(version 11)

New postgresql instances should run on different ports:

root@yeti-postgresql-9:~# pg_lsclusters
Ver Cluster Port Status Owner    Data directory                  Log file
9.4 cdr     5435 online postgres /var/lib/postgresql/9.4/cdr     /var/log/postgresql/postgresql-9.4-cdr.log              <<< OLD CDR database
9.4 routing 5434 online postgres /var/lib/postgresql/9.4/routing /var/log/postgresql/postgresql-9.4-routing.log          <<< OLD routing database
11  cdr     5437 online postgres /var/lib/postgresql/11/cdr      /var/log/postgresql/postgresql-11-cdr.log               <<< NEW CDR database
11  routing 5436 online postgres /var/lib/postgresql/11/routing  /var/log/postgresql/postgresql-11-routing.log           <<< NEW routing database

Shutdown CDR billing process and other pgq-processor consumers

# systemctl stop yeti-cdr-billing@cdr_billing
# systemctl stop yeti-delayed-job

Stop CDR writing to old Postgresql CDR instances

To stop CDR writing change listen_address to ‘127.0.0.1’ at postgresql.conf Configuration for old postgresql instances and restart postgresql

Change epoch of new postgresql CDR instance

Shutdown new postgresql CDR cluster and change epoch:

# su - postgres
# /usr/lib/postgresql/11/bin/pg_resetwal -e 1 /var/lib/postgresql/11/cdr

Where 1 is increased epoch - it should be greater then epoch of old intance

See https://github.com/markokr/skytools/blob/master/doc/faq.txt#L50 for details

Start new postgresql CDD instance.

Copy data from old databases to new instances

You should dump your databases using pg_dump and then restore data to new instances:

root@yeti-postgresql-9:~# pg_dump -h 127.0.0.1 -p 5434 -U yeti -O -x -F c -f yeti.dmp yeti
root@yeti-postgresql-9:~# pg_restore -h 127.0.0.1 -p 5436 -U yeti -d yeti -e yeti.dmp
root@yeti-postgresql-9:~# pg_dump -h 127.0.0.1 -p 5435 -U cdr -O -x -F c -f yeti.dmp yeti
root@yeti-postgresql-9:~# pg_restore -h 127.0.0.1 -p 5437 -U cdr -d yeti -e yeti.dmp

Shutdown old databases and move new instances to their ports

Change port at postgresql.conf of each postgresql cluster to move new instances to old ports. Restart postgresql.

Apply first stage of migrations

root@yeti:/# cd /home/yeti-web
root@yeti:/# su -s /bin/bash yeti-web
yeti-web@demo-yeti:~$ RAILS_ENV=production ./bin/bundle.sh exec rake db:migrate
yeti-web@demo-yeti:~$ RAILS_ENV=production ./bin/bundle.sh exec rake db:second_base:migrate
....SKIPPED...
== 20180212105355 MultipleMatchingConditions: migrated (2.0862s) ==============

IMPORTANT: Now update and restart your servers. And after that run `rake db:migrate` again.
root@yeti:/home/yeti-web#

Restart all components of yeti-web

root@yeti:/# systemctl restart yeti-web
root@yeti:/# systemctl restart yeti-cdr-billing@cdr_billing
root@yeti:/# systemctl restart yeti-delayed-job

Switch to new routing schema

Change at /etc/yeti/system.cfg switch routing schema to switch17 from switch16 on your yeti-management server and restart yeti-management server

SEMS servers

Upgrade your SEMS nodes to new version. Configuration files format was changed, See sems.conf for details.