Is it safe to upgrade Ubuntu from 14.04 to 16.04.1 via do-release-upgrade, and how-to


(Anton) #1

My server tells me:

New release ‘16.04.1 LTS’ available.
Run ‘do-release-upgrade’ to upgrade to it.

Should it be safe to:

  1. Turn off the server.
  2. Make a droplet backup while offline (DigitalOcean).
  3. Run do-release-upgrade

Any things to pay attention to?


(Oleg Bovykin) #2

Just be aware, that restore from backup takes a lot of time. For my 10$ instance it was ~1hr.


(Jay Pfaffman) #3

Maybe easier to create a new droplet and restore to it. You can get a floating ip,assign it to your old droplet until DNS propagates, build the new one and switch over when you like it.


#4

I did the upgrade without any problems on DO a few weeks ago.

what I did:

  1. Live snapshot on DO

  2. stop discourse: ./launcher stop app

  3. upgrade with screen: do-release-upgrade

I believe I had to manually start docker service docker start once the upgrade was done, but didn’t have any issues otherwise.


(Oleg Bovykin) #5

Its actually faster if you create new droplet and set up discourse.
Its just system configuration, docker installation (simple apt-get) and rsync from old instance.


(Anton) #6

Hi Oleg, may you please show an example of working rsync commands for the approach proposed? What folders exactly need to be synced?

Except speed, are there any other benefits of going this way?


(Jay Pfaffman) #7

It’s just the time it takes to bootstrap. I think it’s the case that if you don’t know the rsync command, you don’t care.


(Anton) #8

Just a note that might be vital for someone:

Taking an online snapshot, even with your DigitalOcean droplet OFF, might be pretty time consuming:


(Anton) #9
  • The DigitalOcean snapshot took ~20 minutes for a 40GB instance.
  • don’t forget running ./launcher cleanup before turning off your droplet and making a snapshot - it may a big difference in the snapshot size, sometimes up to 10GB
  • The do-release-upgrade took about 20 minutes for me (upgrading from 14.06 LTS to 16.06 LTS)

A few notes from the upgrade process I run through:


Answered NO.


Answered YES.



At the end:


After restarting:

Didn’t need this in my case.


Something I did not pay too much attention to, but should have done so when upgrading through a SSH connection - only googled about it later:


End of Life for Ubuntu 14.04 Trusty and Discourse 1.9.0.beta10?
Ubuntu unattended security upgrades
#10

@mr8, I just ran the do-release-upgrade upgrade but got disconnected when it asked me to restart. Now I am getting the error message:
Device "docker0" does not exist. Cannot connect to the docker daemon - verify it is running and you have access

How were you able to manually restart it once the upgrade was done? I am not trained as a programmer so everyone’s detailed instruction would be much appreciated!


#11

Running the following command solved the problem:

./launcher stop app
./launcher start app

Thanks everyone!


(Jeff Atwood) #12

I am upgrading a droplet from 14.04 to 16.04 right now using do-release-upgrade and following @meglio’s excellent summary. So far so good!

edit: completed and running as 16.04 without any issues at all.


(Peter N Lewis) #13

I’ve used the same process on a number of Digital Ocean servers, so it is good to hear this works as expected.

One point, it looks like you are on an older Digital Ocean server, so it has not upgraded the Kernel. You can switch (in the Digital Ocean control panel) the Kernel to the Grug 0.2 kernel, and then Ubuntu will use the latest version (4.4 something).

BTW, the Snapshot is definitely essential - I just tried this process and it corrupted the harddisk and refused to boot properly - I’m lucky like that. I restored from the snapshot and everything was good again (tried again and the same thing happened so now I’ll see what the DO folks have to suggest).

So kids, always do the Snapshot (ideally from a powered off state, event if that means your server is down for an hour).


(Peter N Lewis) #14

I’m with @meglio, I’d like to see instructions on transferring a forum from one instance to a new one.


(Jeff Atwood) #15

Those instructions already exist in the #howto area unless you insist that the transfer has to be rsync, versus simple web download of backup, web upload of backup, then web restore.


(Peter N Lewis) #16

Turns out my fstab had an incorrect UUID for the root disk which was the root cause (get it! ;- ) of the resulting disk corruption. Fixing that meant the update to Ubuntu 16 worked without further problems.


(Justin Gordon) #17

Any idea how much free disk space is needed to do this?


(Anton) #18

I had at least 7GB free, but don’t know how many are needed actually.