DataStore Backup Location for Standby Machine

[FredSpataro]

Hi All, 

I've added a backup/standby machine to the my site's hosted datastore... Everything looks good but by default value for the standby machine's backup location is on the local machine directory.  Should this be changed to the same folder as the network share for the primary?  Or will cause an issue?  Should it be a different location on the network share? Just leave it local while it's the standby, then change if it ever becomes the primary? Something else? 

Thanks! 

[CodyPatterson]

Hey @FredSpataro 

It may not be entirely best practice, but I basically have a sister share that exists in the same location as the primary backup location if space allows. So the folder structure goes //holding_folder/backup_primary and //holding_folder/backup_secondary. In terms of backup location, I don't believe there's any standard best practices on where exactly to put it, as the backup for the backup has not so far been used in my environment. We've utilized the standby machine around 15 times through various power failures or environmental issues.

Cody

[FredSpataro]

Thanks Cody... I like your approach...seems like a 'standby' shouldn't even run a backup but also doesn't seem to make sense to have the location same as the primary in case something goes wrong or conflicts... so a different location covers the bases.

[FredSpataro]

Update:  Interesting result from adding a machine in two environments (TEST & PROD)... 

• In my test environment, the new standby datastore loaded with the default values for backup schedule and location ie not the same values as the primary
• In my prod environment, the new standby datastore loaded with the same value for both backup schedule and location as the primary

The one notable difference is that in TEST, I installed the datastore, configured it, then ran the patch utility which found a 'Highly Available' patch which I installed --- but after the store/machine was joined.  In PROD, i knew the patch was needed so I installed that patch right after installing the software, before doing the configuration wizard.  All suspect, as I swear these never seem to go exactly the same ever... The patch doc doesn't specifically say it, but does seem like there was an issue with backups:  

• BUG-000156440 - Restoring highly available relational ArcGIS Data Store through webgisdr shows PostgreSQL databases return primary statuses for both machines when point-in-time recovery (PITR) is enabled.

So maybe having the standby configured exactly the same is the correct setup?