Difference between revisions of "Failsafe"
From ArmadeusWiki
(→Generate it) |
(→Restore eMMC from backup) |
||
| Line 35: | Line 35: | ||
<pre class="apf"> | <pre class="apf"> | ||
# gunzip -c /mnt/nfs/emmc.bak.gz | pv | dd of=/dev/mmcblk2 conv=sync,noerror bs=512 count=7733248 | # gunzip -c /mnt/nfs/emmc.bak.gz | pv | dd of=/dev/mmcblk2 conv=sync,noerror bs=512 count=7733248 | ||
| + | </pre> | ||
| + | |||
| + | ===Backup rootfs partition only=== | ||
| + | * Check that you are using the right partition ;-) and be sure that it has no errors (here example for OPOS6UL(NANO)): | ||
| + | <pre class="apf"> | ||
| + | # dumpe2fs -h /dev/mmcblk0p2 | grep name | ||
| + | # e2fsck /dev/mmcblk0p2 | ||
| + | </pre> | ||
| + | * Mount rootfs in a temporary mount point: | ||
| + | <pre class="apf"> | ||
| + | # mkdir -p /tmp/mnt/rootfs | ||
| + | # mount /dev/mmcblk0p2 /tmp/mnt/rootfs/ | ||
| + | </pre> | ||
| + | * eventually get current date: | ||
| + | <pre class="apf"> | ||
| + | # udhcpc -i eth0 | ||
| + | # ntpd -nqp 217.147.208.1 | ||
| + | # DATE=`date +%Y.%m.%H-%M` | ||
| + | </pre> | ||
| + | * backup will be put on HOME partition so check there is enough space: | ||
| + | <pre class="apf"> | ||
| + | # df -h | ||
| + | </pre> | ||
| + | * and launch backup: | ||
| + | <pre class="apf"> | ||
| + | # cd /tmp/mnt/rootfs/; tar cvzf /home/rootfs_$DATE.backup.tar.gz * | ||
</pre> | </pre> | ||
==Links== | ==Links== | ||
* | * | ||
Revision as of 16:33, 25 July 2018
Contents
Intro
Failsafe is the short name for the RAM bootable rescue image for APF boards (APF6 & OPOS6UL only for the moment). It is a Linux kernel + minimal initrd image all packed together in a standard kernel image (apf6-linux.bin or opos6ul-linux.bin).
Installation
Generate it
$ make apf6failsafe_defconfig or $ make opos6ulfailsafe_defconfig $ make $ cp buildroot/output/images/*-linux.bin /tftpboot/
Launch it
BIOS> run ramboot
Cool stuff that can be done with it
apf-config
# apf-config
- check eMMC filesystem
Backup eMMC
- you need a network connection with a server that can receive the backup (for example with NFS, here mounted on /mnt/nfs/):
# dd if=/dev/mmcblk2 bs=512 count=7733248 conv=noerror,sync | pv | gzip -c -9 > /mnt/nfs/emmc.bak.gz
- wait some minutes until backup is done ;-)
Restore eMMC from backup
- same principle:
# gunzip -c /mnt/nfs/emmc.bak.gz | pv | dd of=/dev/mmcblk2 conv=sync,noerror bs=512 count=7733248
Backup rootfs partition only
- Check that you are using the right partition ;-) and be sure that it has no errors (here example for OPOS6UL(NANO)):
# dumpe2fs -h /dev/mmcblk0p2 | grep name # e2fsck /dev/mmcblk0p2
- Mount rootfs in a temporary mount point:
# mkdir -p /tmp/mnt/rootfs # mount /dev/mmcblk0p2 /tmp/mnt/rootfs/
- eventually get current date:
# udhcpc -i eth0 # ntpd -nqp 217.147.208.1 # DATE=`date +%Y.%m.%H-%M`
- backup will be put on HOME partition so check there is enough space:
# df -h
- and launch backup:
# cd /tmp/mnt/rootfs/; tar cvzf /home/rootfs_$DATE.backup.tar.gz *
