Redundancy setup for Rundeck community edition

[JM]

Rundeck community,

Does anyone have experience setting up a Rundeck community edition with some sort of redundancy? Active-active with automated switch over is ideal, but active-passive with a manual switch-over is acceptable. 

Thank you.

Jim.

[rac...@rundeck.com]

Hi Jim,

That is available on Process Automation products, take a look.

Regards.

[Jin Mao]

Right, I see that the commercial version has a nice cluster feature. 

Will there be any issue if two Rundeck community servers access the same database to behave like a simple active-active setup? 

If not possible, I am thinking of having one standby server with the same configuration files for an active-passive setup. In case the primary server is down and not recoverable, the standby server will be powered on and take over the DNS name to resume the service with a minimized downtime. Does it make sense?

Thanks!

--
You received this message because you are subscribed to the Google Groups "rundeck-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rundeck-discu...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/rundeck-discuss/e6bae00f-b103-4ffc-8ebf-ff3abbe891e2n%40googlegroups.com.

[Rubal Chawla]

Hey

I have tried Active-Passive setup in past using Pacemaker and Corosync. Only drawback is maintenance and if failover happen when a job is running you have to manually restart the job.

You can give it try for minimal clustering and failover.

Regards

[Ben Kaplan]

you can also use other tools for similar outcome like AWS ASG which will do health check and automatically replace the instance (if running on EC2).

you will need external DB (RDS) and keep the job outputs also externally (S3) so the instance will only hold the service and it's configuration.

On Mon, Oct 16, 2023, 19:46 Rubal Chawla <ruba...@gmail.com> wrote:

Hey

I have tried Active-Passive setup in past using Pacemaker and Corosync. Only drawback is maintenance and if failover happen when a job is running you have to manually restart the job.

You can give it try for minimal clustering and failover.

Regards

On Monday, 16 October 2023 at 19:39:17 UTC+5:30 JM wrote:

Right, I see that the commercial version has a nice cluster feature. 

Will there be any issue if two Rundeck community servers access the same database to behave like a simple active-active setup? 

If not possible, I am thinking of having one standby server with the same configuration files for an active-passive setup. In case the primary server is down and not recoverable, the standby server will be powered on and take over the DNS name to resume the service with a minimized downtime. Does it make sense?

Thanks!

On Mon, Oct 16, 2023 at 10:00 AM http://rac...@rundeck.com <http://rac...@rundeck.com> wrote:

Hi Jim,

That is available on Process Automation products, take a look.

Regards.

On Monday, October 16, 2023 at 10:55:34 AM UTC-3 JM wrote:

Rundeck community,

Does anyone have experience setting up a Rundeck community edition with some sort of redundancy? Active-active with automated switch over is ideal, but active-passive with a manual switch-over is acceptable. 

Thank you.

Jim.

--
You received this message because you are subscribed to the Google Groups "rundeck-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rundeck-discu...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/rundeck-discuss/e6bae00f-b103-4ffc-8ebf-ff3abbe891e2n%40googlegroups.com.

--
You received this message because you are subscribed to the Google Groups "rundeck-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rundeck-discu...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/rundeck-discuss/64349ed8-5367-4e3d-a015-41535fcd84can%40googlegroups.com.

Disclaimer

The information contained in this communication from the sender is confidential. It is intended solely for use by the recipient and others authorized to receive it. If you are not the recipient, you are hereby notified that any disclosure, copying, distribution or taking action in relation of the contents of this information is strictly prohibited and may be unlawful.

[Roger McCarrick]

Jim,

the answer is yes, I do it.

I have 2 Rundeck front end Redhat linux servers.
We access with FDQN (https://rundeck.mycompany.com) through a Netscaler. So you can hit one or the other. If one is down, the Netscaler knows and takes it out.

Both Rundeck's point to the same SQL database, which is a pair of SQL servers in an availability group.
Both Rundeck's point to the same active directory. Users login with their AD credentials.
Both Rundeck's have the same DFS share mounted and point to it for their logs.
Some config files and ACL files are rsynched from one to the other daily.

I have cascading options that use json files on the Rundeck server, they too are rsynched.

So no matter which front end the Netscaler sends you to, you can see all the jobs and all their logs.
And you can see all scheduled jobs from either one.

My Rundeck jobs mainly launch powershell scripts on a windows node.
I have 2 Windows nodes for redundancy . Now this is a little tricky. The projects are usually built with a single node specified.
It may be possible to have 2 nodes, NODE-A and NODE-B and if Rundeck cannot connect to the first, try the other. I don't know how to do that.

So what I did what make up a virtual node name, VNODE. That points to the IP of NODE-A in the /etc/hosts file on each Rundeck server.
In the projects, I add VNODE instead of the real name of either node.

There is a bash script in cron on both Rundecks that detects if the nodes are down, run it every X amount of minutes. 

If NODE-A goes down, it updates /etc/hosts to point VNODE to NODE-B.
When NODE-A is back up, it switches back to NODE-A.

Scripts are in github, so they are pushed to both nodes. But u could replicate files with robocopy.
We have tested shutting down nodes and Rundeck servers ... and all is good.


Roger