Hi, I am planning to use c6i or above nitro instances for a long time deployment. What should I expect on maintenance from AWS side. Do VMs are getting rebooted on a regular basis for maintenance or is there any case of maintenance for them? Does AWS notify me on such case?
AWS won’t automatically reboot VMs for you - that would be very troublesome for production deployments. Maintenance comes down to a couple things:
• ensuring your disk doesn’t fill up, which is highly dependent on what you’re running on the instance. If you’re not writing anything to disk then you don’t have to clean up the disk much. If you’re writing to disk a lot, then you’ll need some scripts that will clean out directories regularly. You can set alarms on volume storage too so you know if the disk if filling up
• security updates - you’ll want to have a cron job or something that installs any updates for your OS and software
If possible, the preferred pattern is to regularly recreate the instance, usually when you deploy. This mitigates both the concerns described above (though you may still need to manage the disk depending on how frequently you’re deploying and how much you’re writing to disk)
You could schedule something to automatically recreate your instance every so often as well
Not for me but I though if they have to reboot instances in case of hardware failures or bios updates etc.
Never seen it personally, I’ve had instances run for years without being rebooted. Wouldn’t recommend it, but I think it’s extraordinarily uncommon to lose an on-demand instance.
I have experienced occasional reboots due to hardware failures. AWS will send advanced notice if they can, it still requires a reboot but you can plan the time. Typically they give a week’s notice. On rare occasions there can also be unexpected hardware failures that can interrupt an instance.
It’s rare, but always good to have a plan for unexpected service interruptions.
