Example:
1. Every Friday evening when our VM full backup runs, it has a partial failure
2. At first glance, it appears only three, seemingly consistent VMs, actually fail
3. At first glance, the failure appears to always be related to VSS wait timeout
4. The failures succeed on the next retry; but I would like to determine the root problem so that they don't fail to begin with
(I always try to fully investigate, identify, and resolve the issue, whether its one of our clients' backup jobs or one of our own, before involving Veeam support.)
To perform an pattern analysis:
1. I create a spreadsheet with the first column representing the VM names having failures
2. Then each subsequent column is a "column pair" where A is the date of the full backup and B is the timestamp of that VMs failure within the job
3. I manually examine each full backup instance going back 2 or more months and notate the results
4. The spreadsheet will then reveal to me (1) which VMs are consistently or mostly consistently failing and (2) if the issue might be time concentric
Figure 1: Showing the pattern analysis sheet and the remaining cells I have to populate with timestamps, which is currently a tedious process
Current process steps:
1. Scroll to line item in job log window
2. Right-click line item
3. Click Copy
4. Select unneeded cell in spreadsheet
5. CTRL+V
6. Select desired cell in spreadsheet
7. Read and type the timestamp from unneeded cell
8. Delete unneeded cell content
In my case, I had to do these steps over and over to capture 17 timestamps; this resulted in 136 steps in total (this does not count moving to the next full backup job log item in the job log list).
It would be really advantageous to have the following capabilities in the job log window:
- Add an optional timestamp column to the left or right end -- this would eliminate having to right-click > Copy > Paste; instead I could just read the timestamp as I type it directly into my spreadsheet
- Filter the rows so that I can show only errors or only warnings -- this would eliminate having to scroll unnecessarily
The two capabilities above would essentially result in a "hands free" process after opening the initial job log:
(assuming filters and timestamp preferences were persistent)
1. Read the timestamps and type directly into the desired cell in spreadsheet
In my case, this would have resulted in 17 x 1 = 17 total steps to accomplish the same thing in my spreadsheet.
Now, going one step beyond...
I would also be really beneficial to filter the actual job log list which currently can filter on job name through the Search field. True dynamic filtering, however, would make each column filterable. Microsoft does this in Excel with checkboxes when you convert the view to a table. BackupRadar does when they essentially put a Search field at the top of most columns so you can literally "create your own filter set" to achieve the outcome desired.
Figure 3: Showing BackupRadar's filtering system which works amazingly well and is super flexible
Adding this capability into Veeam would then allow me to filter the job log list on Failure and then "step" through the list by using the left and right keyboard arrows from the job log window.
Going one final step beyond...
Since BNR already knows everything that has happened (full historical knowledge), it could perform the pattern analysis for me and provide a summary:
- Over the past six months...
- The Friday full backup initially fails to backup the following three VMs...
- The common failure is: VSSControl / Unfreeze error
- The failures fall between 10:00PM and 2AM (there are no deviations outside this range)
- Occasionally one or two other VMs fail with no consistent pattern