Administrative, Technology and Society

OK, New Problem

25 Comments

I’m going to take the drive to a recovery service in Studio City this afternoon to get an estimate. Meanwhile, I want to boot back into my regular OS, but I doubt if I can do it without sdb1, because the fstab specifies that it overmount /home at boot. So I either have to set up one of the other drives as /home (not sure how to do that, other than mounting it and adding myself as a user) and make it sdb1, or get into the fstab on the SSD to tell it not to do the overmount. I haven’t been able to figure out how to read or write to sda3, which is my Fedora partition, to get at etc/fstab. Any ideas?

[Update a while later]

OK, I’m now booted into my regular system, with one of the new drives as /home. But since I currently have no data, there’s a lot of rebuilding and reconfiguring to do.

Off to Studio City now to drop off the drives.

25 thoughts on “OK, New Problem”

  1. Remove all drives except the SSD and reboot. Login as root and cd into /etc. As root you should be able to edit the fstab file. But honestly you don’t need to do that. From your stick OS remove all drives except SSD and ONE of your new drives and reboot. With the old sdb drive gone your new disk should appear there instead. Make sure you can mount it and can see the lost+found directory. Then shutdown and boot up normally on your SSD. It should automount ok but you may get errors you’ll have to straighten out later. No matter. After the system is up login again as root. Then create a new user for yourself. That should allow you to login as yourself minus your old data. Once that’s working you can reattach the second new drive and reboot. I have no clue how lvm will deal with that. Probably will attach them together as one virtual drive. Hopefully it will ask first. Not an LVM guru. I avoid it. Suggest you read up on backup strategies for lvm volumes. Or make it a separate drive.

    1. I don’t think you understand the problem. It’s very likely that the SSD won’t boot without /dev/sdb1, and I can’t figure out how to get to /etc on it from the stick OS to change the fstab. When I try to mount it, I get: mount: /mnt: unknown filesystem type ‘LVM2_member’.

      1. Great ok. Well try it anyway first with just the SSD and if that doesn’t work with the SSD and one of the new drives. If just the SSD boots ok you should be able to login as root and edit the /etc/fstab file. But I’m thinking you should try to boot the SSD with just ONE of the new drives connected and see if it will come up. If it doesn’t the problem probably has to do with LVM. Your new disk should configure as /dev/sdb with the old sdb disk removed. On reboot everything will move up to take the old drive’s place.

        LVM stitching disks together behind the back. No thanks. I can live without.

        Anyway try booting off the SSD w/o any other drives connected and let me know what happens. If it doesn’t work fine. As we have already done we can burn that bridge when we come to it.

  2. I should say the /etc on the SSD. So mount the SSD first and then cd into the /etc on it. But again you should need to do this. See above.

  5. You’re welcome. I’ve been recovering from linux problems like this professionally since the late 90’s. Last time was just 2 weeks ago.

    I thought your user’s home directory would probably be there, but in case it wasn’t, I thought it best to make sure you could login from the graphical interface.

    I know from experience how bad it sucks to accidentally reformat the volume you are trying to recover.

    In the future, remember to keep your backups current, either to the cloud or to a USB drive.

    1. I’ve been recovering from linux problems like this professionally since the late 90’s.

      Wow. I knew Linux had problems, but didn’t realize people had to recover from OS disasters so often that they could spend 20 years going it in the pro leagues! ^_^

  6. Rand, for future reference the steps you needed were:

    1) parted or fdisk /dev/sdd and /dev/sde one partition to be the size of disk, type Linux.
    2) ddrescue -f -n /dev/sdb /dev/sdd /root/ddrescue-sdb2sdd.log
    3) e2fsck or fsck /dev/sdd1
    4) mount /dev/sdd1 /home
    5) /bin/df /dev/sdd1 should match what it says for sdb1
    6) ls /home and poke around, should be ok.
    7) umount /home
    8) ddrescue -f -n /dev/sdd /dev/sde /root/ddrescue-sdd2sde.log
    9) fsck /dev/sde1 if you are totally paranoid. should not be necessary.

    Charles, Peter, MichiCanuk check me if I’m wrong but I don’t think it was necessary to do a mkfs.ext4 on either of the new drives since they were being cloned by ddrescue. Just partition as Linux device. Usually disk mfgs, init the partition table to type NTFS (ie Windows, used to be FAT32 but not on a 2TB drive) so that the partition shows up by default under the Windows XYZ disk format utility. If ddrescue fails complaining about the output then you can format the output device.

    Pull all drives except SSD and the one new disk. You can boot into single user mode on the SSD to check to make sure the new disk configures correctly as sdb1 (with no other drives except the SSD at sda it should). Then since it is a clone copy, LVM ought to be happy with it so, reboot the SSD normally and see if you are back. If new the drive refuses to configure at sdb then in single user mode edit /etc/fstab to mount the drive at whatever it configures as, with partition 1. i.e. /dev/sd1 Keep the other drive for safe keeping.

    Charles can you recommend some good reading for Rand about backing up LVM disks? LVM is also not in my wheel house.

    1. Oops WordPress grabbed my angle brackets: that should read as: i.e. /dev/sd(c,d,e…)1.

    2. I’m not familiar with ddrescue, but programs like clonezilla and G4L copy everything, including MBR, partition information, i.e., everything. So no formatting is needed as the copy program includes all of the formatting. So I would recommend skipping any formatting, especially for the SOURCE. It would not matter for the destination, as the copy program will destroy it anyway.

  7. Rand, I had some time this afternoon so I contacted Western Digital customer support. As it turns out neither your old drive or your new ones feature any type of write protect, either in hardware or firmware. So I don’t feel so guilty not telling you about a non-extant feature. I recommend they pass along to the design team the idea of modifying the SMART standard to include a mechanism for write protecting the drive at least at the firmware level. But the short form of this story, there was no way to configure your hardware to protect against this mishap. There was a time when you could. Many external flash drives do. Welcome to 2021.

  8. None of this makes me want to install Linux anytime soon. My PC is a tool, not a hobby.

    1. Linux is an excellent choice for older machines and reliability is fine if you don’t live on the bleeding edge. I’ve had fewer problems with Ubuntu than with the various incarnations of Windows. And if something catastrophic does happen, you are much more likely to recover with Linux (with a lot of free software and help) than you get with MS. I’ve known people who throw up their hands and buy a new laptop when they can’t get their Win machine to work. A colleague of mine had a real “hangar queen” corporate laptop, always being “fixed” by IT. Don’t get me started on Win Enterprise! AAARRRRGGGGGGHHHH!!! How many times do you have to go for a coffee (and danish) when you try to reboot after having been afflicted with an Win10 update? I use Win when I must (preferably in a virtual machine running within an Ubuntu host), but I prefer Linux, which is in fact much easier to install.

      1. One more thing. There is an annoying “feature” of many Linux distributions that is a result of the OS’s modularity. When you upgrade to a new version of the OS, sometimes an old library gets deprecated and it breaks an application. I recently experienced this when the latest long term support version of Ubuntu (Focal-Fossa) was incompatible with a library needed by a graphics program that I’m very fond of. Fortunately, the newest short term support version fixed this problem, so I upgraded a few years before I normally would. Upgrading was rapid and uneventful. I’ll need to upgrade from Groovy-Gorilla in a year or so, but then I should get back to a long term support version. If you want a more static and stable environment, one should consider Debian. Although Fedora’s user interface has improved since things have gone Gnomish, I still don’t like its maintenance features.

        If you like to be able to customize your working environment, Linux has no equal.

        1. I haven’t gotten there yet but in my perfect world Linux is running on my rack servers and Windows is just a VM running under Linux. Then when Windows does what Windows tends to want to do, reconstruction is a trivial copy and VM restart.

          Unfortunately, for the time being it is just ass backwards. Running Ubuntu 18.04 on a VM under Win 7. I’m afraid I have an affliction for Win 7. I have used both Win 7 and Win 10. For my home machines, which are Win 7, I don’t see the point of a newer downgrade. I keep my anti-virus up-to-date.
          Office 2013 works just fine as well. I don’t need Microsoft holding my files for ransom either. Everything else is Linux.

          I’m pretty particular about Windows. After severe burns from Win 95, I have only used (in succession) Windows 2000, Windows XP and Windows 7. Nothing else and for good reason. I never had to suffer Me, Vista and the various Win 8 atrocities which could not decide if the GUI should be that of a tablet or a desktop on a docked laptop PC. Thank you no. I don’t need a schizoid OS. Nor do I desire to keep a minimum of 16GB of RAM available just to boot.

          1. I started as an OS/2 devotee, because in the old days, you could do machine control, network, and multitask with it much better that any Win until later versions of XP. For work, I had to use Win for administrative tasks, but OS/2 save my bacon for many years, being able to multitask while running true-blue DOS machines with legacy device drivers for ancient hardware.

            I have experienced ME (horrid) and the laptop I’m posting this on came with Vista (bloated, but not as bad as Win8). I had to buy an upgrade to Win7, but Win10 refused to install, with the world’s most useless error messages!! Even the log files were useless. So I moved over to Ubuntu and am running the most up-to-date version on it (20.10 Groovy Gorilla). The machine is more sprightly than when it was new.

            I’ve got Ubuntu and Mint virtual machines on my newer Win10 laptop, and I’ve got Win10 in a Virtual machine hosted by Ubuntu 20.10. I much prefer Linux as a host to having Win as a host. With Linux as a host, Win10 can magically read all sorts of file systems that it otherwise couldn’t (Ha Ha).

          2. Back when I had satellite Internet service (no DSL or cable available at the time) I installed the one-way satellite receiver in a Dell mini and installed a third party package to run the satellite receiver card and dialup modem uplink. Effectively turning the Dell into a satellite router. The third party software only ran under Linux. I used RedHat 7.2. All my other Win PCs connected via their Ethernet ports as if I had a cable or DSL modem / router. It worked really well. Maximum uptime I got on that system between reboots was slightly over 240 days of continuous operation.

  9. I’m going to take the drive to a recovery service in Studio City this afternoon to get an estimate.

    What’d they say? Is the price reasonable?

  10. Pingback: VirtualBox Issues | Transterrestrial Musings

Comments are closed.