Execute commands serially on a list of hosts. Each ssh connection is made in the background so that if, after five seconds, it hasn't closed, it will be killed and the script will go on to the next system. Maybe there's an easier way to set a timeout in the ssh options...

Cleaned up and silent with &>/dev/null at the end. Show Sample Output

Show Sample Output

This will write to TAPE (LTO3-4 in my case) a backup of files/folders. Could be changed to write to DVD/Blueray. Go to the directory where you want to write the output files : cd /bklogs Enter a name in bkname="Backup1", enter folders/files in tobk="/home /var/www". It will create a tar and write it to the tape drive on /dev/nst0. In the process, it will 1) generate a sha512 sum of the tar to $bkname.sha512; so you can validate that your data is intact 2) generate a filelist of the content of the tar with filesize to $bkname.lst 3) buffer the tar file to prevent shoe-shining the tape (I use 4GB for lto3(80mb/sec), 8gb for lto4 (120mb/sec), 3Tb usb3 disks support those speed, else I use 3x2tb raidz. 4) show buffer in/out speed and used space in the buffer 5) show progress bar with time approximation using pv ADD : To eject the tape : ; sleep 75; mt-st -f /dev/nst0 rewoffl TODO: 1) When using old tapes, if the buffer is full and the drive slows down, it means the tape is old and would need to be replaced instead of wiping it and recycling it for an other backup. Logging where and when it slows down could provide good information on the wear of the tape. I don't know how to get that information from the mbuffer output and to trigger a "This tape slowed down X times at Y1gb, Y2gb, Y3gb down to Zmb/s for a total of 30sec. It would be wise to replace this tape next time you want to write to it." 2) Fix filesize approximation 3) Save all the output to $bkname.log with progress update being new lines. (any one have an idea?) 4) Support spanning on multiple tape. 5) Replace tar format with something else (dar?); looking at xar right now (https://code.google.com/p/xar/), xml metadata could contain per file checksum, compression algorithm (bzip2, xv, gzip), gnupg encryption, thumbnail, videopreview, image EXIF... But that's an other project. TIP: 1) You can specify the width of the progressbar of pv. If its longer than the terminal, line refresh will be written to new lines. That way you can see if there was speed slowdown during writing. 2) Remove the v in tar argument cvf to prevent listing all files added to the archive. 3) You can get tarsum (http://www.guyrutenberg.com/2009/04/29/tarsum-02-a-read-only-version-of-tarsum/) and add >(tarsum --checksum sha256 > $bkname_list.sha256) after the tee to generate checksums of individual files !

perl version of "Wait for file to stop changing" When "FileName" has not been changed for last 10 seconds, then print "DONE" "10" in "(stat)[10]" means ctime. One have other options like atime, mtime and others. http://perldoc.perl.org/functions/stat.html

Recent versions of dd provides status flag. Setting that flag to progress shows progress in realtime.

I find the other timers are inaccurate. It takes some microseconds to perform the date function. Therefore, using date/time math to calculate the time for us results in millisecond accuracy. This is tailored to the BusyBox date function. May need to change things around for GNU date function. Show Sample Output

Ummmm.. Saw that gem on some dead-head hippies VW bus at phish this summer.. It's actually one of my favorite ways of using bash, very clean. It shows what you can do with the cool advanced features like job control, redirection, combining commands that don't wait for each other, and the thing I like the most is the use of the ( ) to make this process heirarchy below, which comes in very handy when using fifos for adding optimization to your scripts or commands with similar acrobatics. F UID PID PPID WCHAN RSS PSR CMD 1 gplovr 30667 1 wait 1324 1 -bash 0 gplovr 30672 30667 - 516 3 \_ sleep 3 1 gplovr 30669 1 wait 1324 1 -bash 0 gplovr 30673 30669 - 516 0 \_ sleep 5 1 gplovr 30671 1 wait 1324 1 -bash 0 gplovr 30674 30671 - 516 1 \_ sleep 7 Show Sample Output

It suspends to RAM: you always need your batteries for the RAM but it saves time as there is no need to slowly archive everything on your hard disk. It works fine with me but if anyone has a nicer way, please contribute.