[ale] Buwahahah!! Success!

Thu Aug 31 22:15:56 EDT 2023

Still have to put the data on the disk somewhere! I do use bind mounts when something changes and a chunk needs to be visible somewhere else. Or a dir in a partition needs to be read only.

But lvextend is live so it's real time.

On August 31, 2023 5:45:55 PM EDT, Steve Litt via Ale <ale at ale.org> wrote:
>What's your opinion of using bind mounts to create realtime
>stretchable/shrinkable partitions as opposed to LVM?
>
>Jim Kinney via Ale said on Wed, 30 Aug 2023 20:47:05 -0400
>
>>Lvm is a total lifesaver!! You never know really how to partition a
>>drive so lvm can help expand a partition. On. The. Fly! Add a new
>>drive or add a new raid box and lvm says, sure, let's use that!
>>
>>It also supports software raid which irritates the hardware purists
>>with deeper pockets that me. A software raid10 is cheap, fast,
>>reliable, and if I really need it, I can clone the box into a new mobo
>>with some boot magic and it resurrects the added blank drives in old
>>and new boxes for me without a pair of cards that cost more than the 4
>>new drives. Spinning rust sata drives with 5 year warranties are
>>totally worth it.
>>
>>Yeah. Lvextend is a lifesaver. Lvreduce is awesome as long as the
>>filesystem is not xfs. Ext4 supports shrink. ZFS of course replaces
>>ext4 and raid and lvm but does eat more CPU in Linux land. Pretty sure
>>ZFS borders on being a filesystem cult but the prophets have some
>>really good points. Maybe one day it'll get into the mainline kernel.
>>Probably right after gluster. 😁
>>
>>I'm almost embarrassed to admit that I no longer fix my home gear. If
>>it pukes, it just gets replaced. Hardware is mostly pretty reliable
>>(not gonna discuss HPC/supercomputers running a hot tub style liquid
>>cooling solution). There's used Dell/Supermicro server gear in Suwanee
>>data centers that hits eBay. It's usually 5-7 years old and lasts
>>another 3-5 years in the home shop. 3 on the Supermicro, 5 on the
>>Dell. But at $350 for a dual CPU, 8-12 core, 64-128G ram, add your own
>>hard drives, I'm happy.
>>
>>I do need to kick the backups again. Long overdue for the bare metal
>>recovery of the entire backup system. Thanks for the reminder of "aging
>>backups".
>>
>>On Wed, Aug 30, 2023, 5:36 PM Charles Shapiro <hooterpincher at gmail.com>
>>wrote:
>>
>>> About three weeks ago piglet, my primary desktop computer, pooped
>>> out. Press the power button and the fans came on, but nothing else
>>> happened -- no POST, no screen, like, Nuthin'.  Went through all the
>>> hardware troubleshooting I knew, carted it around to a couple of
>>> friends who are smarter than me, but never revived it. It was a Core
>>> I7 motherboard obtained surplus 5 years ago after a hard life as a
>>> server, so I reckon it was no big surprise it finally bit the dust.
>>>
>>> $500 or so and a couple of sessions at Decatur Makers later I'd
>>> replaced everything but the Mass Storage, the video card, and the
>>> case.  She would boot to the BIOS screen np. I could get the GRUB
>>> screen but no further -- she'd would just Kernel Panic.  The new
>>> guts are a 12th gen Intel I9 on a Gigabyte Aorus Z690 gen 1.4 MB, so
>>> maybes that had something to do with it.
>>>
>>> Fortunately, I keep my OS on a 120 GB SSD, and my /home on a much
>>> larger Spinning Rust drive. So I knew that I wouldn't have to go
>>> back to my (shamefully aged) backups.  I installed Debian 12 on the
>>> SSD (up from Debian 11) and got her to boot ok.
>>>
>>> I configured my original install to use lvm without really
>>> understanding what that meant, so  my /home wouldn't actually, like,
>>> mount with a simple mount(8) command. Cue a deep-dive into lvm,
>>> helped along by an excellent tutorial (
>>> https://linuxhandbook.com/lvm-guide/ ) which also let me delve into
>>> the Wonderful World of Vagrant.
>>>
>>> After groveling through all that mess, I did the following:
>>>
>>> * vgrename the old piglet-vg vgroup to piglet-home-vg ( using the
>>> UUID grabbed from vgdisplay so I was sure to rename the correct one)
>>> * vgchange -ay piglet-home-vg to 'activate' my renamed vgroup
>>> * vgscan --mknodes to fiddle the file system to recognize my new
>>> logical volumes
>>> * Verify that I could now mount(8) my piglet-home-vg/home lvolume on
>>> /mnt (Yay!)
>>> * systemctl set-default multi-user.target to bring the machine up
>>> with no GUI and log in as root
>>>  * Move the installed /home to /home-debian12-default ( in case I
>>> needed to grab some stuff from there to make the Debian 11 settings
>>> for Plasma work with Debian 12).  Make a new empty /home to serve as
>>> a mount point.
>>>   * Edit /etc/fstab to mount /dev/mapper/piglet--home--vg-home on
>>> /home
>>>   * systemctl set-default graphical.target to bring the machine back
>>> up
>>>
>>> Of course I still have a bunch of software to install and some stuff
>>> to bring back from my backup ( all my local apache stuff is gone for
>>> example). But it's really all over but the shouting.
>>>
>>> Fun Things I Learned:
>>>
>>>   * If you screw up an entry in /etc/fstab, Debian 12 will halt
>>> during the boot process when it tries to mount disks.  On some
>>> occasions, it'll attempt to mount your screw up for a while and time
>>> out after a minute and a half or so, but other times I think it just
>>> dies.  You can fix this by choosing Emergency Mode from the GRUB
>>> menu and fixing the bad edit in your /etc/fstab.  Or I suppose you
>>> could boot from your stick again if that rocks your sox.
>>>
>>>   * Debian 12 doesn't appear to let you mount an lvolume from fstab
>>> by UUID. I could do this on my VM, which was running Ubuntu. On
>>> Debian you mount from /dev/mapper, which seems to be the Correct Way
>>> (at least that's the way shipped lvolumes are mounted).  There's
>>> some magic going on here that I still don't fully understand. Some
>>> of the hyphens in the /dev/mapper lvolume names are doubled, again
>>> for reasons which are inscrutable to me.
>>>
>>>   * Hardware can be Tricky.  If you don't plug in ALL the power
>>> connectors on your MB, it will simply refuse to start at all.  Then
>>> you will tear your hair out until you figure out the dumb misteak
>>> you made. And if you get checksum errors late in your install off a
>>> Stick, it means that the media is no good no more.
>>>
>>>    * vagrant and lvm are pretty way kewl.  Learning on a virtual
>>> machine let me hack away at lvm and other scary stuff (like
>>> parted(8) and mkfs(8) ) break things, and still not disturb anything
>>> important on my personal machines.  Highly recommended.
>>>
>>> All in all a lot of fun.
>>>
>>> -- CHS
>>>
>>>  
>
>
>SteveT
>
>Steve Litt 
>Autumn 2022 featured book: Thriving in Tough Times
>http://www.troubleshooters.com/bookstore/thrive.htm
>_______________________________________________
>Ale mailing list
>Ale at ale.org
>https://mail.ale.org/mailman/listinfo/ale
>See JOBS, ANNOUNCE and SCHOOLS lists at
>http://mail.ale.org/mailman/listinfo
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.ale.org/pipermail/ale/attachments/20230831/d237bb0c/attachment.htm>