Add the ability to create automatic backup schedules, much like Rackspace and others, for persistent disks. Currently the only way we have to do something like this is to run a cron task from an instance with gcloud commands. Preferably, we'd be able to take snapshots on regular intervals from a simple admin interface instead. Bonus points for adding rolling last n backup capability, like - last three hours, last three days, last three weeks, last three months.
I’m excited to announce a new capability for GCE, scheduled snapshots for persistent disks, is now available in beta. To learn more about how to use this new functionality, please visit the announcement page (https://cloud.google.com/blog/products/compute/introducing-scheduled-snapshots-for-compute-engine-persistent-disk).
Gavin Bainbridge commented
Great new feature, but please put in an edit schedule button. I forgot to add the right label so I have to recreate the schedule and apply to the disks again.
Christian Wallentin commented
Please add "Monthly" as a schedule as well. Snapshots is a great way to get "longer" backupperiods without much hassle. I have previously done a snap roughly once a month and kept them for 6 months. (Too much hassle to get one box to handle all ComputeEngine in terms of access to manually for now)
With Monthly setting I would do a backup on the 1st of each month and keep it for 6 months.
In addition, setting a number of snaps to keep would also be useful. Even setting the time to keep using the same unit as the backup schedule, ie if it is daily the unit to set how long to keep the snapshots should be in days as well, not hours.
AdminAri Liberman (pm, Google) commented
Apologies for the inconvenience. We identified an issue that required projects to be added to a whitelist. We've rolled out a fix to all regions that removes this requirement and snapshot schedules are now working.
https://issuetracker.google.com/issues/124654199 - you can follow the availability fixing in this issue
Etienne Lessard commented
The function doesn't seem to be creating daily snapshots at all.
PCS Web commented
I've followed the directions given, but it doesn't seem to be creating daily snapshots at all.
We would love to see this feature ASAP along with bonus points.
Philip Brown commented
Get on this please. We've been waiting too long
Levi Koenig commented
This needs to be added! The hacked together methods on GitHub that allow people to accomplish this show that there is demand for this. I have a project and I am not overly technical. The snap-shot functionality gives me some relief knowing that I have access to my project in case something goes wrong. Currently I have to go into the interface everyday and manually create all my needed snapshots. This is time consuming and annoying.
Gourav Kapoor commented
I believe we dont only need to have the scheduler for automatic snapshots but we should have following features built in.
1. Snapshot Groups -- Where I could mention no. of Disks or group specific disks together to be backed up at same time)
2. Snapshots Via Labels - Where snapshot should only occur for certain machines or disks based on labels.
Bob Riker commented
How is this not a feature? Please add it ASAP!
Rob Archibald commented
Would love to see this on GKE as well!
Hi, we evaluate our feature requests backlog continuously with each release, and internally at Google we do planning approximately quarterly. "Under Review" here essentially means it is tracked and we're evaluating it, gathering requirements, and studying the market need. We don't state something is "Planned" until we have high confidence that engineering is going to start working on it. So, in this forum, it may be possible to see suggestions "Under Review" for quite some time, unfortunately.
In the case of this feature, we're actively investigating requirements with users that potentially need this feature.
When will this be reviewed again? It's been 2 years since the last "under review" status
NIc Boyde commented
Presently cron jobs cannot be run on GC Engine - some footling objection to disappearing instances etc - but a cron-job that limits itself, say, to creating snapshots of all available instances need not worry about instances that aren't there - just back up those that aren't.
this will be a MUST feature if GCP wants to have any big Enterprise customer. I am going to have number of GCEs in multiple projects. I am not sure how I can reliably/consistently manage the snapshots of many GCEs without the automatic backup schedules service.
I hope you implement this, it´s hard to keep fresh snapshots manually.
This is the only feature that I think is missing on Computer Engine. I have other providers and offer automatic snapshots: las 24 hours, last week and las 15 days. Having this here will be great and help us to decide us to move everything to Google.
Clayton Gulick commented
Hi Scott - agreed, though for the specific use case of automating backup schedules a simple admin UI might encourage adoption, though I like the flexibility of the more general GCP cron service approach.
This would be a great feature alright. To me, it seems like the best approach would be to provide a general, cross-platform solution, like a GCP cron service, so that it can be used for this use case, as well as use cases for other products (e.g., expire/delete GCS objects/buckets, periodically run a dataflow job, etc.). Thoughts?