Any potential issues with (even temporarily) relocating the cb folder out of /usr?

Bastille

Verified User
Joined
Mar 25, 2013
Messages
97
Due to some brilliant decision making on my part when setting up the partition sizes on some of our DA servers, the /usr partition is only 5GB in size on them versus the 10GB on our other servers. As a result, these servers can't update PHP or MySQL because they're too close to the disk space limit. I was able to frankenstein a solution for one by moving and then symlinking the php directories inside CB while it was updating them but that's not a desirable solution to perform every time.

In a perfect world, I could just shut the servers down temporarily and then run GParted to fix things. Since they're VMs, I could probably rely on a snapshot as a backup in case things do go awry. Given the potential problems from a worst case scenario though, I'd prefer to leave it as a last resort approach.

While messing around with the mickey mouse'd symlink "fix", I noticed that CB will remove symlinks if they're already present before you update as well as after you update. I'm probably being overly cautious but want to make sure that moving the CB directory elsewhere and then symlinking it back into place won't cause problems as well.

Additionally if I'm missing some viable alternative solution besides symlinking or shutting down briefly to muck about with the partitions, I'm all ears.
 
Hello,

Yes, I saw the same custombuild does not like to be under a link. In this case I could suggest that you create a symlink for /usr/local/directadmin/, instead of /usr/local/directadmin/custombuild/

Note, I did not test it on my end and can not guarantee it will work as desired. At least it's worth a try, isn't it.


Or use mount --bind.
 
If you're using LVM:

1) And have free space in your volume group, you can extend your /usr partition, or
2) Create a separate partition for /usr/local, and move all the data over there, then mount your new partition.
3) If you don't have free space, you could add another disk to your VM, extend the volume group, and then 1) or 2)

If you're not using LVM, then Alex' suggestion about using a bind mount might be your best bet.
 
Unfortunately only one server is using LVM and that one isn't one of the affected servers.

I will give mount --bind a shot. One of the affected servers is our testing server so if something does go wrong, it's not a big deal.

Thank you for the recommendation.

Edit:

Yep, that seems to have done the trick. Now I just need to mount it in fstab to make this permanent.
 
Last edited:
Back
Top