I had a trouble with a long process running and wish to know how much I/O this process is doing, so I wrote this quick and dirty python 2.x script: import time,sys,datetime def read_stat(pid): f=open("/proc/%s/io" % pid ,"r") for line in f: if line.startswith('rchar'): rchar=line.split(':')[1] continue if line.startswith('wchar'): wchar=line.split(':')[1] continue f.close() return int(rchar),int(wchar) pid=sys.argv[1] r0,w0 = read_stat(pid) while 1: time.sleep(1) r1,w1 = read_stat(pid) print "%s\t\tr=%s\t\tw=%s" % (datetime.datetime.now().time(),r1-r0,w1-w0) r0,w0=r1,w1 You must give the process PID number as input to the script. In the output you get the read/write throughput of the process in bytes per second, for instance: ...

Starting with a plain old one-disk configuration… # df -h Filesystem Size Used Avail Use% Mounted on /dev/sda2 5.8G 590M 5.0G 11% /data thanks to the power of btrfs, let’s add a second hard disk, with mirrored data AND without unmounting/reformatting! :) also note the different size…. # fdisk -l Disk /dev/sda: 6 GiB, 6442450944 bytes, 12582912 sectors Units: sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes Disklabel type: dos Disk identifier: 0xea97ecdc Device Boot Start End Blocks Id System /dev/sda1 2048 526335 262144 82 Linux swap / Solaris /dev/sda2 * 526336 12582911 6028288 83 Linux Disk /dev/sdb: 4 GiB, 4294967296 bytes, 8388608 sectors Units: sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes … state of the filesystem before the change… ...

It’s always recommended to backup your data for safety, but for safety AND security let’s encrypt your backups! to compress and encrypt with ‘mypassword’: tar -Jcf - directory | openssl aes-256-cbc -salt -k mypassword -out backup.tar.xz.aes to decrypt and decompress: openssl aes-256-cbc -d -salt -k mypassword -in backup.tar.xz.aes | tar -xJ -f - Another trick with the tar command is useful for remote backups: tar -zcvfp - /wwwdata | ssh root@remote.server.com "cat > /backup/wwwdata.tar.gz" ...

For testing or development purposes, I do a wide use of small linux virtual machines. After spawning a new guest (Virtualbox, VMWare or any other), often you want to log on over ssh but you don’t yet know its ip address. You need to login as ‘root’ in the console just to issue a quick ‘ifconfig’, and after writing down the address, you logout and connect with your comfortable terminal. In order to save some time and keystrokes, I put this in my rc.local of all my guest VMs: ...