Installer not creating xfs on sda4?

Project Discussion
1

I’m provisioning with PXE and Matchbox, and it’s not working on a new machine. Error:

:/usr/libexec# journalctl -u ignition-ostree-growfs|cat
-- Logs begin at Tue 2020-05-26 20:50:08 UTC, end at Tue 2020-05-26 20:50:59 UTC. --
May 26 20:50:52 localhost.localdomain systemd[1]: Starting Ignition OSTree: Grow root filesystem...
May 26 20:50:52 localhost.localdomain coreos-growpart[702]: CHANGED: partition=4 start=1050624 old: size=4540383 end=5591007 new: size=15627002511 end=15628053135
May 26 20:50:52 localhost.localdomain coreos-growpart[702]: /usr/libexec/coreos-growpart: line 30: TYPE: unbound variable
May 26 20:50:52 localhost.localdomain systemd[1]: ignition-ostree-growfs.service: Main process exited, code=exited, status=1/FAILURE
May 26 20:50:52 localhost.localdomain systemd[1]: ignition-ostree-growfs.service: Failed with result 'exit-code'.
May 26 20:50:52 localhost.localdomain systemd[1]: Failed to start Ignition OSTree: Grow root filesystem.

This is because /usr/libexec/coreos-growpart is expecting blkid -o export /dev/sda4 to print something that will set TYPE. But it doesn’t:

:/usr/libexec# bash -ex ./coreos-growpart /sysroot
+ set -euo pipefail
+ path=/sysroot
+ shift
++ findmnt -nvr -o SOURCE /sysroot
++ tail -n1
+ src=/dev/sda4
++ findmnt -nvr -o MAJ:MIN /sysroot
++ tail -n1
+ majmin=8:4
++ realpath /sys/dev/block/8:4
+ devpath=/sys/devices/pci0000:00/0000:00:01.0/0000:02:00.0/host10/target10:0:0/10:0:0:0/block/sda/sda4
++ cat /sys/devices/pci0000:00/0000:00:01.0/0000:02:00.0/host10/target10:0:0/10:0:0:0/block/sda/sda4/partition
+ partition=4
++ dirname /sys/devices/pci0000:00/0000:00:01.0/0000:02:00.0/host10/target10:0:0/10:0:0:0/block/sda/sda4
+ parent_path=/sys/devices/pci0000:00/0000:00:01.0/0000:02:00.0/host10/target10:0:0/10:0:0:0/block/sda
++ basename /sys/devices/pci0000:00/0000:00:01.0/0000:02:00.0/host10/target10:0:0/10:0:0:0/block/sda
+ parent_device=/dev/sda
+ growpart /dev/sda 4
NOCHANGE: partition 4 is size 15627002511. it cannot be grown
+ true
++ blkid -o export /dev/sda4
+ eval
+ case "${TYPE}" in
./coreos-growpart: line 30: TYPE: unbound variable
:/usr/libexec# blkid -o export /dev/sda4
:/usr/libexec# blkid
/dev/sda1: LABEL="boot" UUID="1f3e305e-bc3c-4630-ad48-fc16be2872f0" TYPE="ext4" PARTLABEL="boot" PARTUUID="ff72fc91-4b8a-4dfe-84fe-9922f57bde8a"
/dev/sda2: SEC_TYPE="msdos" LABEL_FATBOOT="EFI-SYSTEM" LABEL="EFI-SYSTEM" UUID="307B-73FD" TYPE="vfat" PARTLABEL="EFI-SYSTEM" PARTUUID="7ef8e328-4fd5-410e-8caf-a5133d714ae0"
/dev/sda3: PARTLABEL="BIOS-BOOT" PARTUUID="a2ed3dff-acd1-49ae-a45e-63231fac1823"
:/usr/libexec#

Is this something anyone has seen before? It’s an 8T disk, should be big enough…

At the very least I feel like coreos-growpart ought to set TYPE=unknown before the call to blkid so the error message makes a little more sense.

2

I’m past this, apparently the real issue was that I wasn’t getting a DHCP address. Nevertheless, this error message could be improved by giving $TYPE a default value :slight_smile:

Oops, no, still happening.

[   67.095532] coreos-growpart[738]: CHANGED: partition=4 start=1050624 old: size=4528095 end=5578719 new: size=15627002511 end=15628053135
[   67.148825] coreos-growpart[738]: /usr/libexec/coreos-growpart: line 30: TYPE: unbound variable
[FAILED[   67.168973] systemd[1]: ignition-ostree-growfs.service: Main process exited, code=exited, status=1/FAILURE
] Failed to start Ignition OSTree: Grow root filesystem.
See 'systemctl s[   67.191909] systemd[1]: ignition-ostree-growfs.service: Failed with result 'exit-code'.
tatus ignition-ostree-growfs.service' for details.
[[   67.212851] systemd[1]: Failed to start Ignition OSTree: Grow root filesystem.
DEPEND] Dependency failed for Ignition Complete.
[[   67.233257] systemd[1]: Dependency failed for Ignition Complete.
DEPEND] Dependency failed for Initrd Default Target.
3

hey @neale - thanks for the report. There looks to be at least one bug here. If you could please open an issue in GitHub - coreos/fedora-coreos-tracker: Issue tracker for Fedora CoreOS. It would be great if we could get more information on exactly how you set this system up so we can understand why it doesn’t have an sda4 in the reported output from blkid. We can have the followup discussion on that in the tracker issue.

4

Filed bug 500

1 Like