Hey guys. Im farming chia and some forks on linux ubuntu using gui.
After launching for example 10 forks plus chia my ram usage is about 50%.
After some days usage runs up to 100% and whole farm starts to bug, run slowly.
It makes me reboot whole farm once a week or so.
Is it normal or am I doing something wrong?
Where should I start looking ?
I have 64 gb ram and ryzen 9 5900.
35 disks abut 14 tb each.
Im a Linux newbie tbh
Thanks 
Maybe start cutting some or all of the forks since you can’t farm forks with compressed plots anyway.
Run “top” or “htop”.
That will let you see which processes are using what amount of memory.
There will be no possibility to farm forks with compressed plots ? Could you please give me source of this info ?
This is a simple answer sorry no one has provided any help for you.
so let me first apologize on this forums behalf
This is how Linux is supposed to work.
If u have ram available it will use it especially over time.
this is as designed
theres 2 thing everyone should do in a linux machine for chia imo
To make the change swappiness parameter to be persistent across reboots open the /etc/sysctl.conf file:
sudo nano /etc/sysctl.conf
Locate the vm.swappiness parameter and change its value to 1 or 0. If this parameter does not exist, creat the following line at the bottom of the file.
vm.swappiness=1
A swappiness value of 0 or 1 is recommended for most database servers.
a one liner that should work to increase your swap to 25gb is as follows:
sudo swapoff -a && sudo resize2fs /dev/<partition_name> 25G && sudo mkswap /dev/<partition_name> && sudo swapon -a
Replace <partition_name> with the actual name of your swap partition (e.g. sda5)
((find the name of partition first running a simple)
sudo swapon -a
Note: This assumes that you have free space available on the same disk, and that your swap partition uses the ext2, ext3, or ext4 file system. If it uses a different file system, you will need to use the appropriate tool for resizing that file system.
Note: you may need to first run
sudo apt-get update && sudo apt-get install e2fsprogs
hope this helps people
this will decrease ram usage. and increase your swap usage. curing the problem you face.
further. it may be enough to just change your swappiness value alone.
but if you do both youll be golden
drjones returns
in the fields with the lost chia farmers
the post above may be taken down strictly because people on this forum hate people who like to help others.
so copy to your own notes
peace out
@HavohejPantocrator, I sent you a private message, to caution you about following directions from un-trusted sources.
Point proven.
I wrote this out of the kindness of my heart
Your Only holding yourself back.
Do u listen to a Sense or a trainee
This is extremely basic Linux commands.
And clearly only using built in repositories
When there is legit crimes happening on this site you’ll see no direct maliciousness from me…
More than one person replied, and you focused on one of those people.
I focused on HavohejPantocrator, because he is new to this forum, and he is new to Linux.
Perhaps you (@Jacek) should copy/paste the advice in this thread into the “Tutorial & Guides” section of this forum, for all new Linux users to follow?
I had a very simple question for you, to say exactly what problems you see with what @jonesjr suggested. You have not responded to that, instead as always changed the subject and started attacking me. So, the only way to interpret that is that what @jonesjr wrote is beyond your technical capabilities, yet your only intention was to smear him and you found another target in smearing me. Really nice job.
To answer your questions:
What you do is just smearing people for no reason writing garbage that only exists in your head certifying it as the only gospel out there and trolling the forum looking for long exchanges about parsing sentences and missing semicolons when completely missing the technical side.
As @jonesjr stated:
and after you wrote that smear
I agree with that. There is nothing in @jonesjr post that should trigger the smear that you have there.
I think that you may got mixed up.
You want to increase the swap area, but at the same time you restrict swap usage, at least this is my understanding. Also, to support your statement you use actually the opposite example (db).
I am rather ambivalent about changing swap size for the node. On one hand we have there sqlite, and it would be the best to get it run from RAM, on the other hand, mostly we deal with boxes with small amount or RAM, so we would like to increase it. This is where the problem lies for me.
As far as you example, the reason dbs have swappiness set so low is that those boxes run with tons or RAM, and to speed dbs performance, it is better to have it run from RAM. Therefore, swap is not really recommended, thus as you noted swappiness value is really low (0 means don’t swap, 100 means as much as you can).
At the same time, you are increasing the swap area, so acting against what you just set in that parameter. So, either you set low value for swappiness and decrease the swap area or increase swappiness and also increase the swap area.
At least that would be my understanding of how those two things are working.
No one on this forum reports to you.
Ever since you complained about a user being a troll (@Jacek, tell us who you called a troll), and that the troll should be ignored, and then I commented as to why I replied to the person you called a troll, your comments have been hostile.
In the thread, where you called that user a troll, you wrote:
“That is just standard rambling of a house troll about something that he has no clue about.”
You took issue with others replying to the person you called a troll, and now here we are.
You lashed out in the " How to best utilize disk space" thread.
I replied to you there. You left my questions unanswered (the same accusation you accuse me of).
In that thread, I took the time to address a multitude of your accusations. I will not coerced into a pattern of @Jacek goes on a rant, and @seymour.krelborn replies.
I will follow your own advice, from the thread where you called that user a troll. You wrote:
“He feeds on people replying to his posts, that’s it. Few of us already asked others to ignore his rambling, but apparently for some it is hard to do.”
You also wrote:
“However, pissing contests really don’t work, as that is what he is really good about.”
Going forward, I will follow your advice and ignore your attempts to engage me in a pissing contest.
If youre running farmr, try not doing so, fixed my ram running away in windows.
my method is simply
just to have linux prefer swap over ram… thereby freeing ram.
to swap to ssd, it will be not noticed, and will increase system stability overall.
leaving plenty of ram for on demand applications moving forward. like the chia client itself
than increasing the size of swap itself so it functions like a full system cache to disk.
this is really just a standard recommendation for all databases tho…
there’s just no reason to hold the latest part of the db in ram(which is what its doing) this data can be written to swap than to disk just fine.
because the data is always incoming linux fills ram with database data which is just
pointless. cause once it fills… things get poopy quick, as op described
linux in general should always be tailored to use cases, optimizations most never even consider.
I’m sure there are other far more complicated solutions to get linux to bypass all incoming network data from hitting the ram…
but
it worked for me in the past… flawlessly. is why i share. facing the identical problem as op.
the fact that it fills… is proof that linux itself out of the box is not optimized for chia… there is no use case for having your ram at 100% ever.
imo i guess
I agree with Dr Jones on this one
That is not what I disagreed with. As I said, both approaches have their pros and cons, so it is more like a personal choice and what the box is on hand.
However, for that method (use more swap), my understanding is that the swappiness value should be closer to 100, not to 0. The closer to 0, the less swap is used, the closer to 100, more things go to swap faster. That’s it.
The reason that dbs use 1 or so is that in case swapping will happen, throwing more RAM at the box is a cheap way to avoid it (for high end production cases). Improving on the CPU has usually very little value, and harder to do. RAM is what drives dbs.
What do you mean by “After launching for example 10 forks plus”?
Not to dunk on the Chia mothership, but if Chia itself is a tick buggy at times, how great are 10 clones of said software going to be? I think you’re pointing the right finger.