Although this is not supported it works quite well.
Only problem you will face is the MySQL configuration: portal is using a "lower_case_table_names=1" that breaks API Gateway when you'll try to upgrade but is needed so that portal can works. Below a working my.cfg for both Portal and API Gateway.
The 2 tables DATABASECHANGELOG and DATABASECHANGELOGLOCK have to be recreated lowercase in "ssg" database.
# ll /var/lib/mysq/ssg/database*
databasechangelog.frm
databasechangeloglock.frm
As answered by CA, this is *not* supported and we only use it on our Sandboxes, not in Dev, nor Staging nor Prod.
Also our Sandboxes have twice the RAM in order to cope with both product running concurrently.
[mysqld]
# The default character set is utf8
skip-partition
lower_case_table_names=1
collation-server=utf8_unicode_ci
init-connect='SET NAMES utf8'
character-set-server=utf8
datadir=/var/lib/mysql
socket=/var/lib/mysql/mysql.sock
user=mysql
symbolic-links=0
bulk_insert_buffer_size=0
# using the defaults for the hardware appliance / x86_64 build
innodb_data_file_path=ibdata1:100M:autoextend:max:15851M
innodb_additional_mem_pool_size=2M
innodb_log_buffer_size=1M
innodb_buffer_pool_size=64M
# only need this buffer for the
# mysql tablespace for authentication
key_buffer_size=64M
query_cache_size=32M
query_cache_limit=8M
innodb_log_file_size=16M
# flush binary logs on commits.
innodb_flush_log_at_trx_commit=1
binlog_format=MIXED
max_binlog_size=500M
expire_logs_days=10
# 5 gigs of binary logs, at most. Probably less.
# And sync it too, so that the logs and the files and any remote systems are all
# in agreement. Its a performance hit for reliability.
sync_binlog=16
# set larger connection limit
max_connections=2625
# increase timeout
net_write_timeout=120
low-priority-updates=0
transaction-isolation=READ-COMMITTED
# and ignore connect errors so it retries a lot
max_connect_errors=2000000
max_allowed_packet=1G
interactive_timeout=864000
slow_query_log=1
# Uncommment log-bin, log-slave-update and log_bin_trust_function_creators
# if a clustered db server
#log-bin=/var/lib/mysql/ssgbin-log
#log_bin_trust_function_creators=1
#log-slave-update
# uncomment the next item on 1st db master server
server-id=1
# uncomment the next item on 2nd db master servers
#server-id=2
#relay-log = /var/lib/mysql/ssgrelay-bin
#relay-log-index = /var/lib/mysql/ssgrelay-bin.index
#relay-log-info-file = /var/lib/mysql/ssgrelay-bin.info
# Slave reliability items:
slave-skip-errors=126,1053,1105,1129,1158,1159,1160,1161
slave_compressed_protocol=1
slave-net-timeout=30
slave_exec_mode=IDEMPOTENT
# Two months of logs rollover. Its not size dependent.
# This opens up the risk of having failed replication for 62 days.
# If this happens, then a failed master will lose all data.
# This is only for low traffic sites. For High traffic, 62 days is far too long.
# For high traffic the rotate binlogs is size deterministic, therefore more reliable.
# This can be enabled, but its NOT a good idea.
# expire_logs_days=62
[mysql.server]
user=mysql
basedir=/usr
[mysqld_safe]
log-error=/var/log/mysqld.log
pid-file=/var/run/mysqld/mysqld.pid
[mysql]
# command line imports of audit data can fail with max allowed packet size
# This will help
max_allowed_packet=32M
[client]
default-character-set=utf8
[mysql]
default-character-set=utf8