Volume expansion is now available!

Hey everyone,

We have just introduced a new flyctl command that will allow you to extend the size of a target volume.

I know many of you have been waiting on this feature and it’s now available with version: v0.0.350.

  flyctl volumes extend <id> [flags]

  -a, --app string      Application name
  -c, --config string   Path to application configuration file
  -h, --help            help for extend
  -s, --size int        Target volume size in gigabytes

Global Flags:
  -t, --access-token string   Fly API Access Token
  -j, --json                  json output
      --verbose               verbose output

A few notes about this feature:

  1. Attached VM’s leveraging the target volume will require a restart so the FS can be resized.

  2. This feature can only be used to scale a volume up. Scaling down is not currently supported.


We are playing with the idea of allowing users to define auto-expansion rules for Volumes. If this is something you’re interested in, what kind of options would you like to see?


This doesn’t seem to work.

I resized a volume from 1 GB to 3, flyctl volumes list reports the size as 3GB, trying to re-run the extend says Error failed to extend volume: Invalid volume size - Specified volume size must be greater than 3 so Fly really thinks it is 3 GB.

However, inside the restarted VM I still get

/ # dmesg | grep vdc | head -1
[    0.060596] virtio_blk virtio2: [vdc] 2088960 512-byte logical blocks (1.07 GB/1020 MiB)

/ # df -h /blah
Filesystem                Size      Used Available Use% Mounted on
/dev/vdc                988.0M    902.5M     18.4M  98% /blah

The kernel sees a block device of only 1 GB:

/ # cat /sys/block/vdc/size
/ # resize2fs /dev/vdc
resize2fs 1.46.2 (28-Feb-2021)
The filesystem is already 261120 (4k) blocks long.  Nothing to do!

The volume in question is vol_53q80vdewqk4gzy6.

Hey @Tv1,

I’ve identified the issue and am working on a fix now. I’ll keep you posted.

Thank you for fixing the general problem; don’t worry about my volume. I’m piggybacking on an incompatible wire protocol change (not by choice!) to bring up a second server and migrate to that.

Sounds good! Just as an FYI, this problem should be fixed. If you come across any other issues, please let us know!