What I learned and how I recovered from Partition Table corruption due to Power failure and fsck

Had an interesting experience. Our power blinked (rapid off and on) few times. Not only did SuKam UPS pass that through the power fluctuations but continued after power has stabilized. By the time I switched off the computer, the damage was done. It would only boot to Grub prompt.
Used a SSD drive from another computer to boot the computer and ran fsck. Instead of fixing the situation, it f*ck’ed up the three partitions of /dev/sda (/boot, swap and /) and created a single partition!
I hadn’t saved the partition information and didn’t remember how much I allocated to swap so I couldn’t safely guess their size and re-partition the disk myself.
I tried rescue option of parted and part, none of them helped.
Then I used testdisk from CGSecurity. It accurately detected the partitions and saved it. Then I fired up gparted. It couldn’t read /dev/sda3 but suggested I should reboot first, which I did.
Funnily the system was rebooting with /boot -> /dev/sda1 instead of /dev/sdb1 (SSD added for recovery) but then mounting /dev/sdb3 as / instead of /dev/sda3. This happened because the menu.lst specified the volume by label and both sda3 and sdb3 had identical label:

kernel /vmlinuz-2.6.18-408.el5 ro root=LABEL=/ rhgb quiet

During booting I changed it to use ID: root=ID=yuwew…
The ID used was of /dev/sda3
It came up but gave an error in nvidia.
I decided to try again on the actual system. I removed the extra SSD to allow the actual SSD to boot properly.
It came up fine but nvidia driver was not loading. After working for so loong with multi-monitors, it was strangely restrictive trying to get work done with a single monitor, duplicated. After some frantic search I found that the best option was to download it from nvidia and run it using init 3.

chmod 755 NVIDIA-Linux-x86_64-340.96.run
./NVIDIA-Linux-x86_64-340.96.run

Then switched back to init 5, rebooted and it multi-monitor started working again.
Take away from this experience:

  1. Always backup partition table.
  2. Do not run fsck without backing up partition table.
  3. Don’t expect fsck to always do the right thing. After all the name is intentional and the tool is to be used only as a last ditch effort to save your disk and data.
  4. UPS should not be your only protection. I am thinking of adding a spike buster in between.
  5. Regular backups are a must. I had some backups but when it happened I realized they were much too old to be of much use. Cloud backup services like Dropbox are your friend.
  6. Backup is of no value unless you remember how to restore them in an emergency.
  7. Backup disk on the same machine was unharmed. So backup even on the same machine is of some value.
  8. Next time around I will choose Graphics Card with seamless Linux support
  9. Always have a second computer around in running condition, even a Raspberry Pi is good. You may need lots of help from the Internet.
  10. Do not setup your Internet connection, router, firewall, DNS, DHCP etc. on your machine. Use your router and share with all your machines from there. Your router is less likely to fail than your machine. You may keep the settings on an unused NIC on a computer to use if your router fails.

PS. The SuKam Tubular battery has failed which SuKam support insists as the cause of this catastrophe. It was aged over 5 years.

How to auto-update CNote, Simple CLI Note Taking Application, to Google Drive Cloud

Super simple command line note taking tool for tracking breadcrumb trail during application development…
CNote is a super-simple command line based Note taking application which we use to track changes, take notes before signing-off etc. during development. It can be used on all variants of Unix, Linux and Mac OS. It’s power lies in quickly taking notes as you are working, without interrupting your flow. The notes are searchable and stored in a viewable text file. You can download it from CNote – CLI Note Taking Application site or directly install it from command line with the following:

wget --no-check-certificate -q -O- https://raw.githubusercontent.com/angsuman/cnote/master/install.sh|bash -

CNotes is super-fast because it stores the notes locally, specifically in ~/.notes.
As we started piling up valuable notes you may start thinking what-if you machine crashes?
One simple solution is to create a symbolic link to it from DropBox managed folder. This ensures DropBox updates it in their cloud whenever new notes are added.
DropBox Nazi says: No DropBox for you!
Incidentally I use CentOS 5.1.x (latest update) on Desktop, which is not supported by DropBox. I also use Raspberry Pi, which also is not supported by DropBox.
Enter Google Drive to the rescue
I tried to make DropBox command line tool work but I found out that DropBox command line client will not work without full installation. This is not possible as CentOS 5.1.x and RaspBerry Pi are unsupported platforms.
Google Drive has a nice third-party command line tool called gdrive (download link) which is just right for my purpose. So instead of wasting more time trying to make DropBox work on RaspberryPi etc., I decided to use Google Drive cloud for backup. I used an unused Google Account for this purpose.
Get access to Google Drive

    • Open a browser and login to Google with the account you want to use.
      • Note: This may be done later too.
    • Upload ~/.notes to Google Drive for the first time:
      gdrive upload ~/.notes
      • This will provide an url which you have to open in your browser to allow gdrive to access files and directories in your Google Drive. After allowing you will be given a code which you will have to copy-paste in command line for the first upload to complete
      • After upload it prints a string (something like: 0B8qjyN8TMP9zSmRYQzhNTaBcdEFg) which is the ID of the file. Copy it. This magic key is the identity of your file on Google Drive.

Backup CNote in Google Drive

    • Now create a file named (for example) updateNoteonGDrive and open it in a text editor and paste the saved file ID.
      nohup ~/bin/gdrive update 0B8qjySampleIDQzhNTaBcdEFg ~/.notes >/dev/null 2>&1 &
  • Make this file executable with chmod:
    chmod 700 ~/bin/updateNoteonGDrive
  • Run it by hand to ensure that it properly updates your notes file in Google Drive.
  • Now open ~/.bashrc with a text editor and invoke this after saving your note like:
    n() {
            echo `date "+%Y-%m-%d %H:%M:%S"`'> '"$*" >> ~/.notes
            tail -1 ~/.notes
            updateNoteonGDrive
    }
  • This ensures every note is backed up in Google Drive. The backup process runs in the background so press Enter and continue with your work. Even if it fails because Internet is down, it will update when you write the next note.

Questions:

Barracuda's Copy and CudaDrive online storage services will be discontinued on 1 May 2016

Last August, LaCie announced its Wuala service was to close in November 2015, and now Barracuda has announced its Copy and CudaDrive services will close on 1 May 2016.
Paying customers are no longer being charged, and all users have until that date to ensure they have copies of all their files in another location.
DropBox, Google Drive and Microsoft’s OneDrive have the lion’s share of the cloud storage mind-share and they are going strong.