Last night I noticed my PC was acting different - my desktop seemed slow or jerky. It had been a couple days since I had re-booted and I had done several updates without rebooting so I figured before I spent time trouble-shooting I would reboot. It was dinner time, so I did not return to my office until this morning. On my monitor was about 4 seconds of boot text with "kernel panic" on the last line. The last time I rebooted I had installed 4.13.0-18 kernel and a new video driver after that and this was the result then - so I had booted into the older 4.13.0-17 kernel back then and it worked fine, but I had forgotten all about that. Without worry, I rebooted again and this time selected the -17 kernel. I saw the kernel panic again - now I worried a bit.
If you read any of my other threads you know I'm a huge BTRFS proponent and write many posts about it's benefits. About a month ago, I wrote an auto-snapshot-and-backup script for my server so I didn't have to do any of that manually. The script runs as a cronjob and takes a daily snapshot, preserves the last three snapshots, and once a week sends a backup to two separate backup drives on the server a drive on my desktop PC - all hands off. Serendipitously, couple weeks ago, I modified the script a little and installed in on my desktop PC too.
Along with the three daily snapshots, I have 5 other installs on my system - which all are on the same btrfs file system, no partitioning required - so I booted into Manjaro and rolled back my KDEneon install to the Tuesday snapshot, but it wouldn't boot either. Tried Monday, but also no-go. Finally booted to the Sunday snapshot - and I was back in business.
Each rollback required:
That's it. The whole process to rollback 3 times took less than 10 minutes.
Additionally, every Sunday my system makes an automatic backup to another drive that is also bootable, but I didn't need to go that far this time. I keep two weekly backups and older one are automatically deleted.
The main point here is: Without BTRFS, I might have been doing a re-install this morning. Instead, I went back to work in under 10 minutes and posted this thread. None of this - snapshots, 6 installs to the same partition, automated bootable backups - is possible without BTRFS.
I have decided that, at least on my desktop, a smarter, more realistic snapshot schedule is in order. If I had waited one more day to reboot, the Sunday snapshot wouldn't have been available so easily. However, as luck would have it, Sunday is the backup snapshot day so it was there on the backup drive if I needed it, but I do not want to rely on luck! It is not uncommon for me to go a week or more without rebooting and if I had gone only one more day, I woudl have been restoring from a backup. A more logical approach is needed.
I have decided a daily backup taken at the first successful login after a reboot is a better approach for my desktop. Then I will know for sure that what I'm leaving behind is bootable and my backups are as well. I am going to ponder the best way to accomplish this. Suggestions or ideas are welcome.
Whatever I come up with, I will document it here.
If you read any of my other threads you know I'm a huge BTRFS proponent and write many posts about it's benefits. About a month ago, I wrote an auto-snapshot-and-backup script for my server so I didn't have to do any of that manually. The script runs as a cronjob and takes a daily snapshot, preserves the last three snapshots, and once a week sends a backup to two separate backup drives on the server a drive on my desktop PC - all hands off. Serendipitously, couple weeks ago, I modified the script a little and installed in on my desktop PC too.
Along with the three daily snapshots, I have 5 other installs on my system - which all are on the same btrfs file system, no partitioning required - so I booted into Manjaro and rolled back my KDEneon install to the Tuesday snapshot, but it wouldn't boot either. Tried Monday, but also no-go. Finally booted to the Sunday snapshot - and I was back in business.
Each rollback required:
- Reboot to Manjaro
- Renaming the subvolume that wouldn't boot
- Renaming the next snapshot in the list to the subvolume name used in grub
- Rebooting to the newly renamed subvolume.
That's it. The whole process to rollback 3 times took less than 10 minutes.
Additionally, every Sunday my system makes an automatic backup to another drive that is also bootable, but I didn't need to go that far this time. I keep two weekly backups and older one are automatically deleted.
The main point here is: Without BTRFS, I might have been doing a re-install this morning. Instead, I went back to work in under 10 minutes and posted this thread. None of this - snapshots, 6 installs to the same partition, automated bootable backups - is possible without BTRFS.
I have decided that, at least on my desktop, a smarter, more realistic snapshot schedule is in order. If I had waited one more day to reboot, the Sunday snapshot wouldn't have been available so easily. However, as luck would have it, Sunday is the backup snapshot day so it was there on the backup drive if I needed it, but I do not want to rely on luck! It is not uncommon for me to go a week or more without rebooting and if I had gone only one more day, I woudl have been restoring from a backup. A more logical approach is needed.
I have decided a daily backup taken at the first successful login after a reboot is a better approach for my desktop. Then I will know for sure that what I'm leaving behind is bootable and my backups are as well. I am going to ponder the best way to accomplish this. Suggestions or ideas are welcome.
Whatever I come up with, I will document it here.
Comment