Backups: Difference between revisions

From CCMDB Wiki
Jump to navigation Jump to search
TOstryzniuk (talk | contribs)
m culling: For Tina
m culling: clean-up
Line 13: Line 13:


=== culling ===
=== culling ===
{{discussion}}
Tina has a reminder to manually cull the backups as follows because they had generated >100GB of files. Would be nice to automate eventually, but leaving for now. As agreed with Trish:  
right now this process has generated over 100GB of dated files. We need to decide how to cull. I think I have this elsewhere on this wiki as well...
I can't think of a good automated process for now, so I propose:  
* for a month, keep the latest file each day
* for a month, keep the latest file each day
* for a year, keep on file on days that divide by 10 (eg May 20).  
* for a year, keep on file on days that divide by 10 (eg May 20) and first of month.
* forever, keep the last one per year.  
* forever, keep the last one per year.
 
Trish, is that OK?
*sounds good Tina.  [[User:TOstryzniuk|Trish Ostryzniuk]] 17:29, 2015 May 20 (CDT)


=== TISS28 ===
=== TISS28 ===

Revision as of 10:53, 25 May 2015

see also Category:Backup

Laptop

Centralized data Backup

Centralized_data.mdb and PHI.mdb live on an eHealth server and so are backed up by them (#eHealth file shares). We also created dated snapshot backups every time the data processor pulls data as part of vetting and then pushes it again. For details, see

culling

Tina has a reminder to manually cull the backups as follows because they had generated >100GB of files. Would be nice to automate eventually, but leaving for now. As agreed with Trish:

  • for a month, keep the latest file each day
  • for a year, keep on file on days that divide by 10 (eg May 20) and first of month.
  • forever, keep the last one per year.

TISS28

TISS28 is not part of our backup process right now. OK for now, because backed up by eHealth, but for consistency we should start sending dated copies to M:\ like the rest of our core data.

I was going to add it the push/pull process but then realized that I am not at all sure that Pagasa doesn’t use TISS on one PC while in pull state on the other. So, we can’t be pulling TISS28 at that spot until we have worked out how Pagasa does this.

We talked the other day about changing Pagasa’s setup to use only one PC. This might be the time to re-visit that, too. So, likely Trish, Tina and Pagasa need to talk before I can integrate this.

Scheduled for 2015-May-25

eHealth file shares

All our file shares are backed up by eHealth. To restore any files missing from a file share contact the help desk.