There's pluses and minuses to each install type (Standard vs Docker which is currently unsupported). Be aware that:
- There is no migration script. Once you've installed with one type there is no "conversion". You'll be installing a new server and migrating agents manually if you decide to go another way.
Tactical RMM does not support changing DNS names, so choose your names wisely. If you need to change your DNS name later you will be uninstalling all old agents, and installing a new server and re-deploying agents.
Debian vs Ubuntu¶
|Base RAM Usage||OS|
|80MB||Clean install of Debian|
|300MB||Clean install of Ubuntu|
Because of RAM usage alone, we recommend Debian 11.
Traditional Install - Officially supported¶
- It's a VM/machine. One storage device to backup if you want to do VM based backups.
- You have a backup and restore script.
- Much easier to troubleshoot when things go wrong.
- Faster performance / easier to fine tune and customize to your needs.
- Docker is more complicated in concept: has volumes and images.
- Backup/restore is via Docker methods only.
- Docker has container replication/mirroring options for redundancy/multiple servers.
- NOT officially supported and NOT recommended for production use at the moment unless you are very comfortable managing/troubleshooting docker.
Azure ranks their VM's in Series https://azure.microsoft.com/en-us/pricing/details/virtual-machines/series/.
Tactical RMM will run poorly in CPU limited VMs. So DO NOT use Series A or Series B VMs. You will need at least a Series F or better. Also, make sure there is no IO throttling / IOPS limits for the VM.
The same applies for other big cloud providers that throttle low end VMs.
If you're having more than 200 agents, that's a larger install. You're probably also a business and making (and saving?) money with Tactical RMM, you're supporting the project right?
You should be aware of server defaults like
Default script run time: 120 seconds
Imagine you have 10 check, on 500 computers, running every 120 seconds.
For checks alone, that's 5000 writes every 120 seconds, or 3.6 million database entries created every 24hrs. Default retention on check history is 30 days, so your check history database is probably averaging 108,000,000 records before regular data purges. That's a lot of write-wear on your SSD-based storage layer.
Do you really need your Disk Space check running more than 1-2 times a day? Probably not.
Also consider staggering the times of all your checks, so that each check is naturally spreading the load on your server at more random intervals instead of focusing all checks at exactly the same time.
So in Summary:
- Please support the project, we do need it!
- Adjust script default run intervals
- Don't have checks (and tasks) all run at the same time
If you have questions please contact us at help[at]amidaware.com to discuss looking at your server and configuration, load testing, and giving recommendations.