Question HP Microserver N54L - NAS

Discussion in 'Networking & NAS' started by dafe01, Aug 16, 2017.

  1. dafe01

    dafe01
    Active Member

    Joined:
    Feb 21, 2016
    Messages:
    810
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    46
    Location:
    London
    Ratings:
    +106
    Hi All,

    I have the above Microserver that I have set up as my NAS using Freenas software running on a USB stick with 4 x 500gb HDD for storage. What i would like to do is to replace the HDD's one by one with 2TB drives. I believe that the drives are hot swappable from what i read somewhere, I just want to know whether this means that i can power the Microserver down and swap out 1 drive for 2TB drive and power it up again and not lose any data on the drives? The HDD,s are not configured in a RAID configuration so if i did this would i lose all the data on the HDD's?

    Any advice for a newbie would be really appreciated.


    ps - I have all of the data backed up onto another NAS but would rather avoid copying all of the data onto the Microserver if possible.


    Many thanks
     
  2. bubblegum57

    bubblegum57
    Well-known Member

    Joined:
    Jan 5, 2010
    Messages:
    1,924
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    86
    Ratings:
    +310
    I don't have a micro server, just a synology nas. As soon as I put in a new drive, the nas wants to format it to ext4. so, all data on the drive is wiped.
    It seems Freenas uses zfs, so unless you format the 2TB drives to zfs first. Perhaps using gparted?
    How will the data get on to the new disk?

    for info, i backup my files to an external hard drive, so I have to copy stuff back which had to do recently, when one of my RED drives packed up.
     
  3. dafe01

    dafe01
    Active Member

    Joined:
    Feb 21, 2016
    Messages:
    810
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    46
    Location:
    London
    Ratings:
    +106
    I have this data backed up to another NAS just want to increase my storage capacity.
     
  4. bubblegum57

    bubblegum57
    Well-known Member

    Joined:
    Jan 5, 2010
    Messages:
    1,924
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    86
    Ratings:
    +310
    But you are going to have to.

    I'm confused, you are going to remove a 500gb drive & fit a 2TB, then you ask will you lose any data.
    When you remove the 500gb drive the data will still be on it. If you have data on the 2TB, unless that is zfs, you won't see the data.

    Perhaps I'm just being dense. o_O
     
  5. dafe01

    dafe01
    Active Member

    Joined:
    Feb 21, 2016
    Messages:
    810
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    46
    Location:
    London
    Ratings:
    +106
    No it's just me being unclear. I think I was confused. I now understand that I have to format the drive before putting it in the server. I will have the data on the removed HDD, I assume so will not lose any data. Thanks
     
  6. bubblegum57

    bubblegum57
    Well-known Member

    Joined:
    Jan 5, 2010
    Messages:
    1,924
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    86
    Ratings:
    +310
    I think what I would do, to make copying faster, remove one of the 500gb drives, insert 2TB format, copy from the other 3 500gb drives to the 2TB.

    But I am not sure how to get the data from the removed 500gb drive.

    Just had a thought, if you have the N54 drives configured as one large drive, JBOD, then removing one, means you might not have the data. Because the data will be spread over all 4 500gb disks.

    That is why I have my 4 drives as separate volumes, if one disk fails, I only loose data from that disk.
     
  7. dafe01

    dafe01
    Active Member

    Joined:
    Feb 21, 2016
    Messages:
    810
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    46
    Location:
    London
    Ratings:
    +106
    That's what I was concerned about initially .....losing data over all the drives from replacing one of the drives.
    Maybe I will simply have to replace one drive and start again. Thanks
     
  8. bubblegum57

    bubblegum57
    Well-known Member

    Joined:
    Jan 5, 2010
    Messages:
    1,924
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    86
    Ratings:
    +310
    Stick a 2TB drive in a usb enclosure, copy the data to that, fit another 2tb drive into the server, & copy back,
     
  9. dafe01

    dafe01
    Active Member

    Joined:
    Feb 21, 2016
    Messages:
    810
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    46
    Location:
    London
    Ratings:
    +106
    That could work. Need a USB enclosure.
     
  10. Navvie

    Navvie
    Active Member

    Joined:
    Apr 17, 2007
    Messages:
    938
    Products Owned:
    2
    Products Wanted:
    0
    Trophy Points:
    66
    Location:
    Fareham
    Ratings:
    +335
    I have not done this, I've always just bought more drives and created a new pool. I also don't use FreeNAS, but it's my understanding that what is done is.

    1. replace drive 1 with a bigger drive. The pool of drives will be degraded, ZFS will resilver the new drive and make the pool good again.
    2. Repeat step 1 for each drive.
    3. Once all drives are swapped and the pool resilvered each time, the extra capacity will be available to you.

    Edit:
    Specific ZFS information is here. Increase capacity of FreeBSD ZFS array by replacing disks | Dan's Blog

    You might be better asking/searching in the FreeNAS forums.
     
    • Like Like x 1
    • Agree Agree x 1
    • List
  11. mickevh

    mickevh
    Well-known Member

    Joined:
    Apr 30, 2007
    Messages:
    7,234
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Location:
    West London
    Ratings:
    +1,743
    For clarity - "hot swappable" means you can take the drives out and replace them whilst the server is powered on (and you need an OS that can cope as well as the hardware.) A lot of enterprise kit is such. My HP microserver is most decidedly NOT hot swappable. Check your hardware manual to be sure.

    In a JBOD configuration, the discs are not "one large drive" (unless your OS has done something to make them so, such as storage pools.) In JBOD configuration, each disc is (should) be completely independent of all the others. Thusly, if you remove it, all the data it contains is "gone" from your server (though is should still be intact on the removed HDD.) In JBOD, removing one HDD should not effect any of the others.

    To spread the data over the discs and/or have some form of redundancy so that if a disc dies, you replace it with a new one and the data is automatically reconstituted, needs some form of RAID type technology. But be aware that some versions of RAID don't offer any such redundancy.

    Wiki's article on RAID is worth a read.

    ZFS is not a universal "catch all" - you can create a perfectly legitimate ZFS configuration of JBOD - ie, no redundancy and rebuild capabilities at all. In such a configuration, ZFS resilver won't recover the data. You only get that from ZFS if you (or your OS) have configured ZFS with your discs in a RAID configuration (be it 1, 5, 6, whatever.) ZFS JBOD's are still "just" JBOD's with no redundancy - they just happen to have the ZFS disc format (as opposed to NTFS, ext4, etc. etc.)

    If you just want a "one off" mechanism for connecting up an HDD to read it, you don't necessarily need the expense and a full disc enclosure - you can buy a USB-SATA "lead" (though it actually has some controller chips in it) to connect up to a "bare" drive for very little money (about 12GBP.) You wouldn't want to use such as a permanent solution, but for a one off it's OK.

    There are so many NAS and RAID technologies available now - including various bastardisations - that I suggest it would be unwise to "read across" the mechanisms of one hardware/OS to another. I would echo the thoughts of Navvie that you would be best advised to check on the FreeNAS forums if you are running FreeNAS, and be very clear about what your current disc configuration is (JBOD, RAID, etc.) before you set out. And make sure you have a backup of everything (important) somewhere before you start.
     
    Last edited: Aug 17, 2017
  12. Navvie

    Navvie
    Active Member

    Joined:
    Apr 17, 2007
    Messages:
    938
    Products Owned:
    2
    Products Wanted:
    0
    Trophy Points:
    66
    Location:
    Fareham
    Ratings:
    +335
    Not sure I 100% follow.

    ZFS works best when it can access the drives directly, not through a hardware raid controller. In many cases, this mean a hardware raid controller needs to be configured as "just" a HBA card (presenting the drives attached as Just a Bunch Of Disks), or the ports on a motherboard put into SATA only mode.

    ZFS is both a raid controller and a filesystem. Which lets it do some funky stuff like error detection and, when set up in a Mirror or RAID-Z, correction. No more corrupted file!

    Couple of really good articles on Ars Technica are what got me interested in ZFS and stopped me using hard ware raid controllers.

    I'll link them here. If it's against the rules I understand if they're removed.

    Bitrot and atomic COWs: Inside “next-gen” filesystems

    Ars walkthrough: Using the ZFS next-gen filesystem on Linux

    RAIDZ is still not a back up. :)
     
  13. mickevh

    mickevh
    Well-known Member

    Joined:
    Apr 30, 2007
    Messages:
    7,234
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Location:
    West London
    Ratings:
    +1,743
    I run ZFS under Linux. The reference documentation I use is this one...

    Aaron Toponce : Install ZFS on Debian GNU/Linux

    Well worth a read through if your are interested in how it all hangs together.

    The point I was attempting to make is that just because you've deployed ZFS, you don't necessarily have storage with redundancy (mirror, distributed parity, etc.) unless you have set up your ZFS environment to do so.
     
  14. whitestar96

    whitestar96
    Active Member

    Joined:
    Dec 5, 2006
    Messages:
    582
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    28
    Ratings:
    +63
    If you have modded the bios of the microserver you could put the 2TB drive in the CD bay, then copy the data from one or all of the 500GB disks, then remove them once the copy has completed & put the 2TB drives in the four standard bays.
     
  15. dafe01

    dafe01
    Active Member

    Joined:
    Feb 21, 2016
    Messages:
    810
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    46
    Location:
    London
    Ratings:
    +106
    It has been modded so may consider that as an option. Thabjs
     
  16. Navvie

    Navvie
    Active Member

    Joined:
    Apr 17, 2007
    Messages:
    938
    Products Owned:
    2
    Products Wanted:
    0
    Trophy Points:
    66
    Location:
    Fareham
    Ratings:
    +335
    This is worth doing anyway, even if you then expand the pool with bigger drives, just in case one of the older drives fails during the resilvering process.
     
  17. mickevh

    mickevh
    Well-known Member

    Joined:
    Apr 30, 2007
    Messages:
    7,234
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Location:
    West London
    Ratings:
    +1,743
    He won't get a resilver if the existing discs are configured JBOD - there's nowhere to resilver the data from.

    Maybe it's worth the OP taking a screenshot of his disc configuration and posting it so we know what we're working with.
     
  18. Navvie

    Navvie
    Active Member

    Joined:
    Apr 17, 2007
    Messages:
    938
    Products Owned:
    2
    Products Wanted:
    0
    Trophy Points:
    66
    Location:
    Fareham
    Ratings:
    +335
    Ah, gotcha! I misunderstood. We're using JBOD to describe different things.

    To me, JBOD is a way of describing a collection of disks. As individual discs rather than an array/logical device.

    I think I'm correct to say you mean JBOD to describe a single disk vdev. Which I keep forgetting is an options in ZFS, because - like striped vdevs - there's no parity information to allow ZFS to correct errors or protect against drive failure.

    TL;DR.

    Yes, a screenshot would be useful. :smashin:
     
  19. mickevh

    mickevh
    Well-known Member

    Joined:
    Apr 30, 2007
    Messages:
    7,234
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Location:
    West London
    Ratings:
    +1,743
    Indeed - when using ZFS, one tends to take for granted that we're doing so because we want some sort of redundant configuration and it's easy to overlook that a single device VDEV that just happens to use ZFS as it's file system is possible.

    It begs the question "why would you bother with ZFS if not intending to use the flash stuff?" to which I'd advance the argument that ZFS has a very strong "don't loose any data" ethos and even in a single disc VDEV, it's various checksum mechanisms (which presumably are "scrub-able" - though I've never tried it on a single disc VDEV,) and "copy on write" paradigm are useful, especially if you are interested in making periodic point in time "snapshots."

    I got very interested in snapshots as a hedge against crypto-locking randsomware. I've used my snapshots to "roll back" files to earlier versions a few times and/or recover accidental deletion. ZFS snapshots makes it fairly easy to do so (though not quite as simple as NetWare's "Salvage" function - much missed - sniff :D) without having to dig out a backup.
     
    Last edited: Aug 18, 2017
  20. Navvie

    Navvie
    Active Member

    Joined:
    Apr 17, 2007
    Messages:
    938
    Products Owned:
    2
    Products Wanted:
    0
    Trophy Points:
    66
    Location:
    Fareham
    Ratings:
    +335
    Just FYI, I've just this minute replaced a 3.0TB drive with a 4.0TB one in a RAIDZ2 array, the other three drives are already 4.0TB.

    Procedure was as follows, my pool is named "Pool2" and I'm replacing ada3.

    zpool set autoexpand=on Pool2
    zpool offline Pool2 ada3
    zpool status (to check that the drive has actually been taken offline)
    power off, replace disk, power on
    zpool replace Pool2 ada3
    zpool status (to see that the new drive is replacing ada3 and resilvering has started.)

    Now just waiting for the resilvering to take place. A scrub of this pool takes about 18 hours, I expect resilvering to take a bit longer.

    Edit:
    Resilvering took 26 hours and 13 minutes and auto expanded to 8TB of usable space as expected.
     
    Last edited: Aug 27, 2017

Share This Page

Loading...