pwm

Members
  • Posts

    1682
  • Joined

  • Last visited

  • Days Won

    10

Report Comments posted by pwm

  1. 19 hours ago, Can0nfan said:

    I know this is a big ask to but what about more than 2 parity options??? big arrays like 30 drive and bigger will surely have certain senarios where more than 2 disks fail at one time causing loss of data and unless those of us with huge deep pockets (im a new dad so thats not me lol) would have to create secondary servers or rely on other methods to back these large arrays up in the very rare event of that 3 disk failure senario

    Multiple arrays are great. More than two parity drives is great. But remember that parity is intended for improved availability - it is not a replacement for backup. So you should really, really look into cheap USB disks and copy out your "must not lose" data on them and then store them at parents/children/friends. A single "big oops" with the PSU may kill all drives in the machine. And a fire may kill all data on every piece of hardware in the house. So only use single storage for data you can afford to lose, whatever number of parity drives you may be allowed use of.

  2. 1 hour ago, Dazog said:

    Also tried transfering files via SMB and speeds were 40 times slower.

     

    The out-of-tree driver isn't 40 times slower.

     

    So it's likely to be some option that needs to be tweaked. Possibly some optimization that is default on in the in-tree driver but must be explicitly enabled in the out-of-tree driver.

     

    It's generally better to figure out what option instead of reverting to the in-tree driver, because the out-of-tree driver is most probably the more powerful driver or Intel wouldn't bother investing time with it.

  3. 30 minutes ago, trurl said:

    Some way to force this would save more people than the few with invalid server names. 

     

    It's a separate issue to stop more people from setting invalid names and handle people who already have invalid names.

     

    Same as with location of partition for cache - unRAID can easily stop people from creating new cache drives with wrong partition alignment but the upgrade process must still as early as possible help people who already have wrong align.

  4. 2 minutes ago, Frank1940 said:

     

    But, those who help out those who don't read them, do read them.  It is always helpful to have a reminder of obscure things that cause problems.

    Of course documentation is good to have. I just want to make it clear that lots of people - every day - will manage to get themselves into issues because they do not read documentation.

     

    So best is a solution where they don't have to fallback on other people. For people with notifications turned on, it would cut away a number of support threads if they get a mail telling them the new name - and informing them why the original name couldn't be used. And people who don't set up notifications will have way bigger problems ahead of them.

  5. 5 hours ago, Frank1940 said:

    You should probably consider adding this information in the release notes on every new version (and all of its rc releases) for the next eight stable releases!  That way, 99% of the users will have updated at least once during this sequence!   (And I have not doubt there will be a post from someone five years down the road who finally decided to upgrade...) 

    Lots of users do not read release notes.

     

    But if an auto-rename makes it accessible using the changed name and the user can access the GUI using IP then it would be fine if the GUI clearly indicates that the machine has a changed name - and preferably why the name change.

     

    Edit: With mail support configured, it would be a nice thing to send out a mail informing the user about the name change.