After an upgrade to Parallels Operations Automation (POA) 5.5, when NG servers are rebooted IP addresses are missing from the loopback device.
This may manifest as websites being down:
# curl -k http://website curl: (7) couldn't connect to host
In addition, a telnet to port 80 of an IP of a webserver in the NG cluster times out:
# telnet <IP> 80 Trying <IP>...
In POA 5.5, the
shstg service was replaced with
redis, which does not add exclusive IPs to the loopback device and does not create network scripts like ifcfg-lo.
This has been filed as issue POA-83893, which was fixed in POA 5.5.5. However, servers deployed before 5.5.5 was installed may be affected by the issue.
Please add the configuration files for all exclusive IP addresses.
If there are some manually created configuration files present on the webservers' filesystems, please review them so as not to create inconsistencies.