locked
Removing Disc: Console Unavailable? RRS feed

  • Question

  • Taking KenW's advice to ChkDsk all of the drives in my pool, I tried the .BAT file that iterates through them - but it didn't work.

    Then, having enough free space to support it, I had the bright idea of removing one drive at a time from the pool and running ChkDsk on them one-by-one.

    Started removal of the first disc sometime in mid-afternoon yesteday.

    Now it is close to noon on the following day, and I still cannot open up a Console window - either directly on the box or via RemoteDesktop.

    The Console window flashes momentarily - and appears in the TaskBar - then disappears.

    TaskMan | Performance was showing periods of high (60-100%) CPU Usage but now seems to have settled down to 4-8%.   But no shares are available;  so I'm guessing the removal is still in process.

    This is a 2 tb disc, about 70% full.

    Normal behavior?   Or have I stepped in doo-doo (again...)?

    Wednesday, September 1, 2010 3:30 PM

Answers

    • Shut the server down.
    • Physically remove the drive.
    • Restert the server.
    • Remove the drive (now shown as "missing") via the console.
    • Recover data per this FAQ.
    • There's a chance you'll lose your backup database.

    The time it takes to remove a drive is a result of the failing drive and the way Windows Home Server uses drives. Because data is managed on drives, it has to be managed when a drive is removed. Even a healthy 2 TB drive might take 12-18 hours to remove from the storage pool.

    A large drive farm filled with 250 GB (I assume you meant) drives is not something you want to contemplate. It will use as much power (per drive) as larger drives, so it will consume much more power overall, and you will need many times the number of drive connections, so probably several additional HBAs. And 10x the number of drives almost guarantees you'll be dealing with drive failures regularly.


    I'm not on the WHS team, I just post a lot. :)
    Wednesday, September 1, 2010 6:25 PM
    Moderator

All replies

  • Hi,
    Removing a drive that large may take a lot of time. If it is having trouble it may take even longer (- or never finish?)
    You could check if your server state - is still 'alive':

    • Can you ping it?
    • Can you connect to it using RDP - or even login using monitor and keyboard?
      If so you could try to find out what it is doing (e.g. using taskmanage)

    I can not give you any garantee, but I would go about as follows:

    1. If the server is still like removing a drive (a lot disk activity is a sign): wait...
    2. If it is idle, shut it down.
      If it is unresponsive: power off/hard-reset is your only option then ....

      Restart and see if this gives you control through the cosole.
      If not, physically remove the drive you where trying to remove, then restart. As long as you have folder dulication enable, you should still have all your data. Add a new drive and let duplication do it's work.

    - Theo.


    No home server like Home Server
    Wednesday, September 1, 2010 5:43 PM
    Moderator
    1. If not, physically remove the drive you where trying to remove, then restart. As long as you have folder dulication enable, you should still have all your data. Add a new drive and let duplication do it's work.

     

    You just broached the obvious - which I should have thought of right away: enable duplication for all shares and then just physically remove the drive to do the ChkDsk somewhere else.

    Unfortunately, I turned File Duplication off before doing this - in hopes of speeding up the process.

    Next time around/next disc, I will take that approach.

     

    Does anybody have any idea what happens when I put that ChkDsk'd drive back into the pool?   I'm guessing that WHS just deletes everything on the drive and starts over again.

     

     

    Wednesday, September 1, 2010 5:50 PM
  • > Can you connect to it using RDP - or even login using monitor and keyboard?
    > If so you could try to find out what it is doing (e.g. using taskmanage)

    No problem RDP-ing, but TaskMan | Processes  does not show any particular task as taking that much CPU (in spite of what I said earlier, the box overall is back to using 60-100% CPU).     It must be buried in "System Idle Process".

    I guess I'll just let it run - for a couple days if necessary.   I've got backups, but there is a bunch of transient stuff (TV recordings mostly) that are not up-to-the-minute backed up and  I really don't want to lose a bunch of data.

     

    Losing the server for as long as it takes for the "Remove" process to complete is turning out tb a major drawback to Windows Home Server.   This is *really* hokey.

    Seems like an argument for a physical server case that supports many, many drives and having a bunch of 250-meg drives instead of a lot fewer terabyte or 2-tb drives.   Either that or Windows Server - which, I assume, doesn't take the server down for hours at a time just to swap out a drive..... or does it?

    Wednesday, September 1, 2010 6:04 PM
    • Shut the server down.
    • Physically remove the drive.
    • Restert the server.
    • Remove the drive (now shown as "missing") via the console.
    • Recover data per this FAQ.
    • There's a chance you'll lose your backup database.

    The time it takes to remove a drive is a result of the failing drive and the way Windows Home Server uses drives. Because data is managed on drives, it has to be managed when a drive is removed. Even a healthy 2 TB drive might take 12-18 hours to remove from the storage pool.

    A large drive farm filled with 250 GB (I assume you meant) drives is not something you want to contemplate. It will use as much power (per drive) as larger drives, so it will consume much more power overall, and you will need many times the number of drive connections, so probably several additional HBAs. And 10x the number of drives almost guarantees you'll be dealing with drive failures regularly.


    I'm not on the WHS team, I just post a lot. :)
    Wednesday, September 1, 2010 6:25 PM
    Moderator
  • Well, after 29 hours of downtime and 60-100% CPU usage, it finally came back.

    And with a extra gift, even:  now it's flagging one of the other 2-giggers as "Unhealthy".

     

    My Strategy Going Forward:

    • Turn Folder Duplication back on for all shares and leave it on - always.

    • Set up my backup server to automagically boot itself up at, say, 0500 each morning and run a backup utility that will keep it's shares populated with the same files  as those on the real server, and then shut down.

    • Acquire some sort of utility that can be run to verify that the production and  backup servers contain the same files and email a report to me - and run said  utility after every backup

    • If/when an issue arises that involves removal of a disc, just shut down  the server, yank the disc, and fire it back up in hopes that Folder Duplication  will work as advertised.

    • If Folder Duplication craps out - or, as right now, another disc unexpectedly comes  up "Unhealthy" or just plain dies; fall back on the backup server and resign myself  to losing any files that were created since the last backup.

    • Never, EVER, under any circumstances "Remove" a drive.

    Thursday, September 2, 2010 12:34 AM

    • A large drive farm filled with 250 GB (I assume you meant) drives is not something you want to contemplate. It will use as much power (per drive) as larger drives, so it will consume much more power overall, and you will need many times the number of drive connections, so probably several additional HBAs. And 10x the number of drives almost guarantees you'll be dealing with drive failures regularly.

    The additional electric consumption is what put me off of them in the first place. 

    I will stick with my current larger drives then - subject to my little rant above that came up as the second post in the thread.

    -Is there any consensus on drive sizes?   A sweet spot maybe - like 1tb or 2tb?  

    Or is it just a matter of the biggest drive/smallest number of drives being the best solution?

    Thursday, September 2, 2010 12:38 AM
  • My Strategy Going Forward: 
    Turn Folder Duplication back on for all shares and leave it on - always.  

    Allways, unless the share is holding data you can afford to 'loose' (or have a backup of)

    Set up my backup server to automagically boot itself up at, say, 0500 each morning and run a backup utility that will keep it's shares populated with the same files  as those on the real server, and then shut down. 

    Unless you can afford to loos all your data, I can only advise anyone to keep at least one full backup of you data (preferably off-site). There is allways a chance you loose a file, disk, even even the complete server.

    Acquire some sort of utility that can be run to verify that the production and  backup servers contain the same files and email a report to me - and run said  utility after every backup

    I prefer doing this manually on a weekly basis (or earlier when necesairy) monitoring the backup result.
    Especially as I do not want to backup corrupted data.

    If/when an issue arises that involves removal of a disc, just shut down  the server, yank the disc, and fire it back up in hopes that Folder Duplication  will work as advertised.  

    Allthough this is not usage as designed, it should work. But you will loose your backup database as well as unduplicated data. I prefer using the "drive removal" procedure, unless reality prevents me.

    If Folder Duplication craps out - or, as right now, another disc unexpectedly comes  up "Unhealthy" or just plain dies; fall back on the backup server and resign myself  to losing any files that were created since the last backup.  

    I count this under "server loss". That is when the offsite backup come in. Never run into this situation (on my production server - test/beta envrionments tend to crash now and then - mostly by my own doing)

    Never, EVER, under any circumstances "Remove" a drive.  

    As said: when used in a normal way with "healty drives" the standard drive removal procedure is prefered.

    - Theo.
     


    No home server like Home Server
    Thursday, September 2, 2010 6:59 AM
    Moderator
  • If/when an issue arises that involves removal of a disc, just shut down  the server, yank the disc, and fire it back up in hopes that Folder Duplication  will work as advertised.

    If chkdsk /r on the drive fixes errors, try the removal. If either chkdsk or the removal hangs, shut down, pull the drive, and remove in the console as above. Until the drive has been removed via the console, you may not have access to any files with a shadow on that drive, and files with a shadow on that drive are certainly unprotected by duplication.
    I'm not on the WHS team, I just post a lot. :)
    Thursday, September 2, 2010 12:51 PM
    Moderator
  • > ... I can only advise anyone to keep at least one full backup of you data (preferably off-site). There is allways a chance you loose a file, disk, even even the complete server.

    This is tangental, but are tape backups totally passe'?   My recollection is that a tape could hold a lot more data in a given physical volume than any hard drive - OTOH that was before 1 and 2 terabyte drives.

    With 10-12 TB of data, it seems like portable backup via hard drives is not realistic - or, at least, quite expensive and inconvenient.

    Also, the one backup-to-the-cloud solution I tried (Carbonite) was hopelessly overloaded bandwidth-wise with that kind of volume.

    Thursday, September 2, 2010 6:03 PM
  • Pete, tape isn't affordable for the average consumer.

    To back up very large volumes of data to tape, you will need something like an LTO 5 tape drive @ $2500 or so, plus a cartridge or two (@ $110 ea). For ginormous 20 TB movie collections, an LTO 5 loader (@ $4000+ plus an appropriate number of $110 tape cartridges)) is your best bet as far as tape goes. Backup to tape is, in any case, not supported in Windows Home Server. V1 has issues with backing up tombstones vs. real files, and V2 has Windows Server backup instead of NTBackup.


    I'm not on the WHS team, I just post a lot. :)
    Friday, September 3, 2010 1:10 PM
    Moderator