Author Topic: 18 hour Verify and "handle=5948"  (Read 1709 times)

Offline sonofwatt

  • Full Member
  • ***
  • Posts: 103
  • Karma: +1/-0
    • View Profile
18 hour Verify and "handle=5948"
« on: January 20, 2014, 07:23:24 pm »
Started a verify yesterday. Just checked on it expecting it to be done, and was surprised that nearly 19 hours later, it's still not done.  Is this normal?
QuickValidate usually finishes in under 10 minutes. I'm scared to even attempt a full Verify!

Wondering what the hold up is, I checked the logs.  The FlexRAID.sys.c1-1.log had easily over 100 pages of these lines repeated.
Quote
[2013-12-9 5:42:46.152287][2836][OnSetEndOfFile] \Videos\Temp\[ www.Torrenting.com ] - The.Newsroom.2012.S02E08.720p.HDTV.x264-EVOLVE\The.Newsroom.2012.S02E08.720p.HDTV.x264-EVOLVE.mkv (handle=5948)
[2013-12-9 5:42:46.152289][2836]      [OnSetEndOfFile] (handle=5948) => Exited...
[2013-12-9 5:42:46.152291][1040][OnSetAllocationSize] \Videos\Temp\[ www.Torrenting.com ] - The.Newsroom.2012.S02E08.720p.HDTV.x264-EVOLVE\The.Newsroom.2012.S02E08.720p.HDTV.x264-EVOLVE.mkv (handle=5948)
[2013-12-9 5:42:46.152292][1040]      [OnSetAllocationSize] (handle=5948) => Exited...
What's that about?
It's not part of this verify task, I'm just wondering what it's about.

And looking over at the regular log. All I really see are warnings of new files and a few deleted files.  Nothing unexpected. 

Where else can I look?
« Last Edit: January 20, 2014, 07:35:12 pm by sonofwatt »
Q6600, 8GB RAM, 1PPU, 1URU, 12x1TB DRUs, Server 2012, ReFS, RAID-F

Offline sonofwatt

  • Full Member
  • ***
  • Posts: 103
  • Karma: +1/-0
    • View Profile
Re: 18 hour Verify and "handle=5948"
« Reply #1 on: January 20, 2014, 07:38:12 pm »
I should add that my logs are set to WARN currently.  But they were set to TRACE earlier.  It's possible those lines in the log sample above are from when it was in TRACE mode.
Q6600, 8GB RAM, 1PPU, 1URU, 12x1TB DRUs, Server 2012, ReFS, RAID-F

Offline Brahim

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 8,547
  • Karma: +204/-16
    • View Profile
Re: 18 hour Verify and "handle=5948"
« Reply #2 on: January 20, 2014, 10:08:32 pm »
There is nothing wrong with the logs.
Verify tasks as much time as parity initialization. It is an intensive process.

Offline sonofwatt

  • Full Member
  • ***
  • Posts: 103
  • Karma: +1/-0
    • View Profile
Re: 18 hour Verify and "handle=5948"
« Reply #3 on: January 20, 2014, 11:06:50 pm »
Sorry, I really should have made two posts.

I was curious about those entries and what they indicated, but figured they were unrelated.

The Validate ended up taking 21.5 hours. At the start it was going at around 600MB/s, and finished at 72. Also at the start it was using a lot of CPU cycles.
Brahim, I know update and verify schedules are highly subjective to an individuals usage patterns. But could you give a set of rules for setting them up? There are some pieces of scattered info on the forums, but nothing coherent or complete.  Would you consider doing a wiki entry about this?

From what I've gathered you suggest doing updates in proportion to how often data changes on the array.  e.g. Lots of new files daily, do an update daily. Or a couple files deleted/changed a week, do an update once a week.
And you've suggested to do a validate after a large amount of data has changed. That makes sense.  But I was also thinking, in regards to data rot, there should be a minimum duration between verify/validate runs. Perhaps saying a validate should be run once every 3 months or more often if you're worried/have had data rot in the past.
Also verify/validate, which is the best for detecting data rot?  And with quick validate, does it have the ability to detect data rot? Is there any use with quick-validate, or is it mostly useful for scripting events?

If you can clarify this stuff, I'd be happy to throw it together in a wiki post.
Q6600, 8GB RAM, 1PPU, 1URU, 12x1TB DRUs, Server 2012, ReFS, RAID-F