LHC@home

BOINC LHC is eating too much of my main drive. The new spinning disk should mount a a sub somewhere under root "/." Being a separate drive, will it still exist in the same filesystem so I can create a hardlink to it "from" the existing disk directory being used by BOINC?
1) Where IS that directory?
2) Is there a config file somewhere that lets BOINC know which directory I want it to work in? <--- This is the solution I'd rather use.
3) Is there a single directory for all BOINC LHC (and others?) tasks?

Thanks.

From what I read the solution is quite difficult.
You should run boinc from the OS drive.
Sharing drives is difficult, unless you can install Boinc on the secondary drive.
Best solution? Buy a 128GB SSD, install Linux on it, and run Boinc and LHC from there.

Well, that is A solution but I was hoping for a better solution for this <censored><censored> high $ new system. It MUST serve more than one purpose. LHC ain't the only thing important in the world.

BOINC LHC is eating too much of my main drive.

How much is "too much"?
How much space is left on the drive?
On recent disks the amount of space used by BOINC shouldn't be an issue.

... should mount a a sub somewhere under root "/."

This makes me suspect you are thinking in the wrong direction.
It would make sense to say:
- Create an empty subdirectory under "/" or "/arbitrary_sub/" as mountpoint
- mount the new drive to this mountpoint

Being a separate drive, will it still exist in the same filesystem

It can be mounted (nearly) anywhere to the directory tree but as it is on a separate device it must have it's own filesystem.

... I can create a hardlink ...

No, you can't.
Hardlinks can only be created for files that are located on the same filesystem.

... the existing disk directory being used by BOINC?
1) Where IS that directory?
2) Is there a config file somewhere that lets BOINC know which directory I want it to work in? <--- This is the solution I'd rather use.
3) Is there a single directory for all BOINC LHC (and others?) tasks?

BOINC uses a starting point called "working directory".
It's location can be found in the BOINC messages when you restart your client.
The working directory can be changed to an arbitrary directory but it must be ensured your client has the correct access rights.
Parts of the BOINC working tree (that's everything below the working directory) can be mounted from other filesystems/devices.
LHC@home fully accepts this but other BOINC projects might fail, e.g. primegrid.

computezrmle,

Yes, I try to speak in as high a level as I can when describing tech problems in order to ask questions with the fewest words. Yes, I am aware of most of the linux fs details but not all of course. I am aware of how to set up and mount new drives. :)

"Too much" = too much for me and I am concerned with the number of writes on an nvme drive. BOINC is obeying the limits I set but I had to increase them.


"BOINC uses a starting point called "working directory".
It's location can be found in the BOINC messages when you restart your client."
Where are those BOINC messages to which you refer? I've never seen them. I don't know what "restart your client" means, either. Do you, perhaps, start your BOINC inside the vm window?

"Parts of the BOINC working tree (that's everything below the working directory) can be mounted from other filesystems/devices."
How do I tell BOINC where that working tree IS?

Thanks.

The usual and easy way to setup your BOINC client, attach it to projects etc. would be via the BOINC manager which is a graphical UI and part of the installation package.
Did you ever use BOINC manager?
If not it might be a good idea to make yourself familiar with it as it provides a menue command below "Tools" to view the messages I mentioned.


Since BOINC client can either be installed from a linux distribution repository, from an installer provided by Berkeley or a self compiled package the working dir locations can vary.

There are a couple of methods to find out which path is used.
The hardcore but universal way could be:

lsof |grep "/stderr\.txt"

Found an entry in the very beginning of the log file.
/var/lib/boinc-client (I think it was.)

I shoulda put /var in a separate partition I guess. Still could and just put it on the new drive.

Since /var contains lots of files used by a running system it should not be transferred to a new location from within this running system.
Might be better to stop BOINC and copy only the tree below /var/lib/boinc-client to the new location.
Keep /var/lib/boinc-client as future mountpoint, modify the entries in /etc/fstab accordingly and restart your computer.
If the mounts succeed start the BOINC client.

All work will be accomplished from booting off a live USB.
Including making a backup image of the entire nvme drive, the only one running at the moment.

All work will be accomplished from booting off a live USB.
Including making a backup ...

+1
That's the way it has to be done.