Hi,
Please give me any comments criticisms. Appreciated.
Just a few comments from a quick read through:
- The code you have to upgrade to jessie (presumably from wheezy) isn't very safe (for instance, what if the server needs custom repositories in its sources.list which the script wipes out?). In the UNIX tradition of doing one thing well, I'd suggest just telling the user they need to upgrade first and exiting if they're still on wheezy. (Sidenote, you can get this information from /etc/debian_version; you don't need to ask the user.)
- > echo -e "\033[1;33m""If you're on a shared/virtual server you can't set the time
This isn't usually true. For example, KVM provides paravirtualized access to the system clock. In either case, running ntpd is best practice on the guest always - I'm not sure what the advantage is in stopping ntpd, running ntpdate and starting ntpd again.
- You use sudo in a lot of places, but you also exit at the top if the script isn't run as root; it isn't necessary to do both.
- Related to the above, it's not a good idea to do things like building packages (particularly those downloaded from the web) as root; I'd recommend having your script run as an unprivileged user and using sudo for e.g., invoking apt-get or the final `sudo make install` step.
- Given that your script is intended for jessie hosts, it'd be nice to have bitcoind run under systemd (put a unit file for the bitcoind service under /etc/systemd); use of rc.local is definitely deprecated. Likewise, using systemd would eliminate the need to set up logrotate and would have the added advantage of all daemon logs being in the same place (journald).
- For safety reasons, all complex scripts like this should start with `set -e` - that way if a particular command fails the script will stop executing, rather than continuing in an inconsistent state and possibly breaking something.
HTH, Sharif