Author Topic: Log errors when Ignore Read Error is set to True  (Read 2084 times)

Offline JeffW

  • Newbie
  • *
  • Posts: 36
  • Karma: +0/-1
    • View Profile
Log errors when Ignore Read Error is set to True
« on: January 04, 2013, 07:11:45 am »
On a snapshot raid, when the Ignore Read Error set to TRUE, there should be (1) an option to log the errors and (2) an option to abort the update or (re)build after a user specified number of files with read errors.

Presumably, there could be a low number of files with errors (say 1 - 3)  in an otherwise robust system.  Also, a file may have a large number of errors clustered together so that is why the bad file count should be tracked rather than simply an error count.

This is primarily relevant for large systems that take more than a day to complete an update or build.  (for example I have 6 DRU 1 PPU snapshot raid with 18 TB of data that takes about 36 hours to update on an i7 running Windows 7)  Aborting partway through is very disruptive when the balance of the data is undamaged.  Completing the parity calculation but reporting the error(s) allows for corrective action to be taken while maintaining the integrity of the rest of the parity data.


Offline Brahim

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 8,547
  • Karma: +204/-16
    • View Profile
Re: Log errors when Ignore Read Error is set to True
« Reply #1 on: January 05, 2013, 03:17:45 pm »
Errors should not be tolerated.
If you have errors, you should resolve the root cause.

The default is to abort on error. Ignoring errors should only be in one-off situations.
 

Offline JeffW

  • Newbie
  • *
  • Posts: 36
  • Karma: +0/-1
    • View Profile
Re: Log errors when Ignore Read Error is set to True
« Reply #2 on: January 05, 2013, 08:42:26 pm »
Agreed, but it should not be a time consuming iterative process to find the errors.  Getting a list of them in one shot makes the most sense.