Ubuntu 16.04 LTS

[Wayne Workman]

@Joe-Gill said in Ubuntu 16.04 LTS:

I just updated FOG today, so I’ll be on whatever the most current SVN release is.

Then your servers will have a replication loop…

[Joe Gill]

@Wayne-Workman

Please explain… Is that a good thing or a bad thing? I thought updating both was always good practice.

[Wayne Workman]

It’s a bad thing. rc-7 has a replication loop bug. File sizes will never match because 8 bytes is appended in the checks, and it’ll endlessly delete files and re-copy everything as if it needs done. RC-8 fixes it, I’ve already tested it.

[Tom Elliott]

@Joe-Gill It is good practice, however the comparators would always fail, causing other nodes in the group to delete their files (Snapins/Images). It would still replicate the data to the system, but the next check cycle would remove and do it all over again.

So it could be good as the nodes will ALWAYS have current image files, however it’s a massive amount of work which would cause the network to be in use all the time, if anyone’s imaging off a different node, the files would just remove themselves causing the host to stop imaging.

It would also be a lot more taxing on IO on the server as it is constantly writing the to the log over and over (2/3 logs really) which will also cause a taxing CPU problem (More load on the server and above expected norms (Say a quad core system with a load of 135 or more.)

[Joe Gill]

@Wayne-Workman

How does one go about getting to RC-8?

[Wayne Workman]

@Tom-Elliott said in Ubuntu 16.04 LTS:

So it could be good as the nodes will ALWAYS have current image files

It will never be good. People who image using the nodes will have corrupt systems because they image with an incomplete file 99%+ of the time.

[Wayne Workman]

@Joe-Gill said in Ubuntu 16.04 LTS:

@Wayne-Workman

How does one go about getting to RC-8?

In your git repo,

git pull
git checkout working-RC-8
cd bin
./installfog.sh

This reconfigures git to use the branch.

[Tom Elliott]

@Wayne-Workman Why do you say that? The replication tasks know if there’s already replication tasks operating. If it’s still running during the next cycle check, it will not do the delete stuff.

[Joe Gill]

@Wayne-Workman

Do you have a good wiki on setting up a git repo? I haven’t set one up on these servers. I’ve been using SVN.

[Wayne Workman]

@Tom-Elliott Because replication is always happening, the file will most of the time be incomplete, and DEPLOY tasks will always fail. I have pictures of it, they were emailed to me all day by technicians.

[Wayne Workman]

@Joe-Gill said in Ubuntu 16.04 LTS:

@Wayne-Workman

Do you have a good wiki on setting up a git repo? I haven’t set one up on these servers. I’ve been using SVN.

Just follow the git instructions in the “Upgrade to Trunk” wiki article. But append my steps. FYI Tom gave me the steps earlier.

[Tom Elliott]

@Wayne-Workman I mean, the tasks don’t randomly get deleted. They will only delete if the replication from the master has completed.

There’s a 10 minute gap by default for the next replication cycle to start. If you image bewteen the “cool off” period, it should work fine, but once the cycle starts back up it will delete the files.

[Wayne Workman]

@Tom-Elliott case in point.

[Joe Gill]

@Wayne-Workman

This does sound like what my previous version was doing. My deployments failed almost every time. I had to kill my tasks for each system to finish. I’ve used this image enough now that I know it wasn’t the image.

[Joe Gill]

@Wayne-Workman

So I ran the GIT update using the commands you provided and I show the following:

Tue Aug 09, 2016 17:01 pm
Running Version 16
SVN Revision: 5948

Is that correct for RC-8?

[Wayne Workman]

@Joe-Gill Yep. Correct until Tom makes another change to it.