[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Reorganising disk to increase space in /boot



On 2020-05-21 12:02, Phil Reynolds wrote:
I have landed myself with a problem on one of my machines, in that it
has insufficient space on the /boot filesystem to cope well with kernel
upgrades. The installation was done a few releases ago, and there are
four disks, three identical and one larger, configured as follows:

Each disk, partition 1: 192512 sectors. RAID 1. Used as /boot, 62% used.
Each disk, partition 2: 195311616 sectors. RAID 6. LVM - see below
Three small disks, partition 3: 117073920 sectors. RAID 0.
Larger disk, partition 3: 781266969 sectors.

No gaps or unallocated space exists on the disks.

RAID 0 and linear addition of last partition is used as scratch space
at present. This can be dispensed with, if necessary, and in any case
currently stands just 1% used.

LVM setup:

Single vg, holding:

/home, 54684MiB, 1% used.
/, 1512MiB, 27% used.
swap, 3812MiB
/tmp, 15260MiB, 1% used.
/usr, 38144MiB, 12% used.
/var, 77316MiB, 35% used.

No currently unallocated space.

I have no shortage of space but sadly it's in an awkward layout to put
right. I have tried reducing the size of one of the filesystems in the
LVM, moving extents to fill the gap and reducing the PV size, but I am
still left with the problem that I cannot find a way to move the start
of the PV "up" to allow the extra space to be added to the first
partition. I tried a relatively recent Parted Magic, but this did not
have a way of doing this.

Is there a way I can do this, using some kind of live OS (grml
perhaps?), and achieve a more workable /boot size without having to
consider the more drastic option of "nuke and pave"?


So:

1.  4 @ 100 MB partitions in RAID1:

        100 MB for /boot

2.  4 @ 100 GB partitions in RAID6 with LVM:

         3.8 GB for swap
         1.5 GB for root
        15.3 GB for /tmp
        38.1 GB for /usr
        77.3 GB for /var
        54.7 GB for /home

       (190.7 GB total)

3.  3 @ 60 GB and 1 @ 400 GB partitions effectively in JBOD:

       580 GB for /scratch


It appears that you have 3 @ 160 GB HDD's and 1 @ 500 GB HDD (?).


I suggest:

1.  Backup everything.

2. Do a fresh install onto a 16+ GB solid-state device; the higher the quality/ reliability the better (I prefer enterprise desktop 2.5" SATA SSD's):

     1 GB partition for boot

     1 GB partition for swap

    12 GB partition for root

     1+ GB unused space for over-provisioning and device size variances

3. Get another 500 GB HDD. Use md or LVM to mirror the two 500 GB HDD's. Use LVM to divide the mirror as required for /tmp, /usr, /var, /home, and /scratch.

Alternatively, use md or LVM to put the 3 @ 160 GB disks into a 480 GB JBOD. Use md or LVM to mirror the JBOD with the 500 GB HDD. Divide the mirror as above.

4.  Restore data and services.

5.  Add system drive imaging to your disaster plans and procedures.


David


Reply to: