Agent continually restarts

or is online, but not responding

Every 30 seconds

Agent versions <= 2.5.0 contained a bug that, under rare circumstances, would cause the agent to cache an undeliverable message. This would cause the agent to drop it's connection to the cloud, reconnect, fail to send the message and drop it's connection to the cloud again in an infinite loop.

Fix

The cache needs to be cleared and the agent fully restarted. Run the following command via REMOTE SHELL to flush the cache. You may need to run it multiple times in order to have it executed.

rm /var/cache/ardexa/*

Once successful, restart the agent from the MANUAL CONFIG tab. Then upgrade the agent to the latest version.

Every 2 minutes

ArdexaLinux has a "interface management" service that attempts to keep the Ardexa device connected and operational at all times. Older versions of this service used ping (ICMP) to check for a working connection to the internet. If outgoing ICMP is blocked by the local router/firewall, the service will fail to detect an internet connection and it will begin restarting network interfaces in an attempt to revive the connection.

Fix

There are two options: Allow ICMP through the firewall, or upgrade the ardexa-iface-manager to the latest version which uses TCP to test a connection directly to the configured broker. Please contact your Account Manager to organise the upgrade.

Every 15/30 minutes

ArdexaLinux has a "check active" service that attempts to keep the Ardexa agent running by monitoring it's status. Under certain conditions, agent versions <= 2.5.0 will not respond to status requests and check-active will restart the agent assuming it's dead or broken.

Fix

Upgrade to the latest version of the agent. Rebooting the machine can help aleviate the problem. In rare cases it may be necessary to disable check-active, however this is strongly discouraged unless you have other reliable means of accessing the machine, e.g. SSH.

Force the agent into fallback mode

There is a way to get the agent to ignore it's normal configuration and start with only the bare essentials (version >= 2.5.1)

touch /tmp/ardexa_fallback

and then restart the agent. The presence of this file as the agent starts will force it to ignore it's current configuration and cache, and simply connect to the cloud. This option is extremely useful for situations where the agent is failing to respond, but is still online.