This thread went from cron jobs to RackNerd Los Angeles in six turns. Classic.
Automated snapshots with rclone and cron
Right, back to backups. I tried Gatus, pretty but overkill for "did my tar finish." Settled on healthchecks.io free tier, ping from cron, dead man's switch. If it does not hear from you in 24h you get emailed. Simple.
Healthchecks.io is good. I use this with my script from earlier, the rclone ls check. Now I have two checks, maybe too many but I sleep better. In Poland we say "better safe than sorry" but also "why pay for two when one is free." 😂
Polish hosting is underrated and so is Polish internet. I run my backups to OVH Warsaw, not Vultr. Same S3 API, cheaper cold storage, and the DC is actually in Warsaw not Frankfurt pretending.
OVH cold storage is glacier-class retrieval though. You want that snapshot now you wait six hours. Vultr standard is instant, you pay for the egress once. Different tool for different panic level.
In Japan electricity is so expensive I run backups at night for the cheaper rate. My cron is at 2am not 3am. Also I use wasabi not Vultr because no egress fee. The latency to US is bad but at 2am who cares.
Wasabi has minimum 90 day retention on every object. You delete early, you pay. For rotation with --backup-dir this adds up fast if you move files daily. Calculate before you switch.
I redirect my backups to /dev/null and sleep fine. If the data wanted to live it would have replicated itself.
Oracle Cloud free tier + rclone = unlimited backups for zero euros. They have Warsaw DC too. The catch: your account gets terminated randomly with no reason given. I run six instances and pray.
Communities die when data dies but also when the free tier evaporates. I pay for Vultr block storage on my one remaining CS server. £5/month, sleep like a baby, map changes at 4am with zero lag.