Recover deleted records?

695
3
04-14-2023 05:14 PM
LMedeirosUI
Occasional Contributor

Hi All - I'm pretty sure the answer is no, but I have to ask...

When deleting test records (deleting the record with multiple individual repeats), somehow the wrong record was deleted and ACTUAL data was deleted. This was done in the S123 web app by the survey owner (me). I'm honestly still not sure how it happened, but it did.

I tried to recover the SQLite database from the collecting device (attached), but the S123 field app was open and refreshed the database so it reflected the deletion. At least as far as I can tell when trying "fix database" and looking at it on DB Browser for SQLite.

Does ESRI have backups? I do not have any backups in place, but the data was collected on Wednesday (4/12).

I now understand why it's important to (1) get backups into place even for recent data and (2) why the outbox/sent folder may be more important than the inbox for surveys that aren't expecting records to need updating. Womp womp.

Any other suggestions? I'm so mad at myself.

Thanks in advance for any help!

Lea

0 Kudos
3 Replies
ChristopherCounsell
MVP Regular Contributor

ArcGIS Online has no recycling bin for items or deleted features. The focus is on delete prevention and creating backups. We can help a lot with advice on these, but you sound confident and it was an accident...

Your only option is to contact technical support and ask for a recovery. However they only do this in exceptional circumstances, and it's as a geojson file from the cache. So no attachments and it may be a snap of your layer any time in the past 30 days. Hard to assess if they'll entertain your request. I've only seen it for huge, critical mistakes. Not even sure if they want the public to know this is possible 😕

Let us know if you need help with backups or delete protection.

LMedeirosUI
Occasional Contributor

After this happened, I went and created CSV and FGDB backups of data for each survey I have and placed in appropriate folders. I can't do automatic backups because the company I am working with doesn't pay for them (unfortunately). The admin has requested a tech call, but we shall see.

Are there other ways to backup the data? Other than the obvious downloading it to a computer/cloud drive? It would be great to have this automated, but honestly I don't typically delete any data other than test data... this was just a REALLY bad mistake that I think happened because (1) I forgot to delete test data before making the survey live and (2) when I was deleting the test data, I deleted another test record and didn't refresh the web app before selecting the next one. So it looked like I was selecting a test record, but wasn't. I will swear up and down all sorts of things that I selected a test record (every entry was some version of "Test1" so it's super obvious). Definitely won't be deleting anything without hitting that refresh button ever again. 

0 Kudos
ChristopherCounsell
MVP Regular Contributor

Backups:

  • Manual export and download
  • Python or tool (FME) export and download
  • Python scripts or tool (FME) to assess data for change and make updates only

https://www.esri.com/arcgis-blog/products/arcgis-online/data-management/back-up-data-in-arcgis-onlin...

Smaller jobs, 1 or 2 is fine. Larger (more data) jobs, consider assessing features and 'syncing' with a backed-up enterprise geodatabase.

Otherwise:

  • Delete protection
  • Editing settings
  • Sharing (editing access  + Shared Update Groups)
  • User account access - who has admin access, less is better
  • Don't use admin accounts unless necessary. i.e., don't use the account that has privileges to delete everything in your org, if you don't need to, even if you are the admin. Like windows PC and install 'run as admin'
  • Use good content management. Authoritative status, thumbnails, naming, metadata, clean accounts, removal of unused content. Will contribute significantly to avoiding 'accidents' as people leave professional content alone more
  • Backup before deleting content, always, as process

 

0 Kudos