Automic Workload Automation

Hello all,

We are planning to setup a centralized agent on a Linux server and then we will write a job to ssh to multiple servers from there and perform the necessary steps. How effective this idea would be and how much load can this kind of agent could cope up with. Can we install multiple agents on this central server for fail over. Please share your ideas and we aim for this approach because we don't want to install and handle multiple agents.

Thanks,

Srujan.

The tasks you are asking of your central server are very light-weight, and I would not expect any performance issues to arise.  It only needs to pass a request to another server and wait for a response.

It is easy to run dual agents by putting them on different ports.  But I would question the value of doing so.  

Hi Srujan,

if you are using a centalized agent in combination with SSH you have different disadvantages:

• SSH is not reconnect capable. This means that there is no way to see what happens after a network failure.
• Servers cannot be checked in advance if they are active.
• Various out-of-the-box features of Automic are not available, e.g. hostgroups,  using of resources at job/agent level, etc.

I also do not see any advantage to install multiple agents for this porpose on the same server. If you want to implement this I suggest to install two agents on two different servers so that a hardware failure can be compensated. As Peter wrote the jobs are very light-weight and there should be no problem for the Automic agent - but all jobs share the same infrastructure and OS resources like network interface, network sockets, etc.

Greetings from Germany

Tim