When you asked previously and I replied
Once an HDD is filled with plots then fragmentation is a non-issue because there are not constant read/write/delete operations happening. Plots are read-only to the harvester/farmer. An HDD storing plots is not a system disk, so you don’t need to leave any room for the OS.
I cannot find any evidence that filling an HDD to maximum capacity with plots would have any detrimental effects on longevity of the drive or performance.
5 Likes
I Havnt either. Iv just had multiple disk failures in Linux. And the drives where at 100.3%
The drives don’t die. They can be revived. But the data is lost.
Maybe there’s Somthing about xfs. Or Mhddfs I use for fuseing the jbod to one directory.
I get that it’s not a os. I was just giving a basic example of the problems of a full disk.
But there is still io logging that happens along with the harvester reads .I’m sure? Maybe that’s wrong? Seems really in efficient if it doesn’t have some cache on di
Does a harvester not creat any sort of write tho? No log. No checksums. No cached data as to not read from the whole disk?
I find this hard to believe.
Such as in windows when u format a drive in disk manager it asks if u want to enable compression. It’s creating a tiny log that need read write space. And the end user would never know.
Or if it’s any kind of journaling filesystem on a hard drive. That space is needed. And filling to max would inhibit the process. Decreasing performance.
First, you cannot fill a drive beyond 100% of its available space. I don’t know where you came to the conclusion that you could fill an HDD to 100.3% of its capacity.
Second, even with plot calculators it’s damn hard to leave anything less than 3GB of free space.
Third, you’re asking us to do some homework for you (what does a harvester do?) when you are the one making a claim (filling HDD to 99.9% is a terrible idea).
edit: one is able to farm with read-only access to the HDD – so no, nothing needs to be written to disk at all.
7 Likes
I tottaly get that you think plots are exclusively read only.
But there are many factors that determine this. Compression, log, cache, io.
Some you can’t turn off.
Some that won’t even say it’s writing
. But if your computers a computer and it self optimizes. As computers do unless explicit told not to.
Just By creating a tiny log on a drive to could give u 50% better read performance
Than ya ur gonna bottleneck at 99.9 percent
Ya Linux is funny. And like I said. Thies are drives that data lost immediately after I was peeking around to see how full they actually where
There’s no claim here this is common knowledge I thought….
Every dev gets a log based on its format. Unless explicitly told not to. Or something like ext2.
MEANING WHAT.
Oh that there’s a place that gets tiny bits of reads and writes on every disk takes place. Io.
And when u don’t have room on a fricken drive what happens
I don’t know how more simple I can make this
Here’s another video explaining
I don’t know how much more I can beat a dead horse.
Cmon people
And another brilliant explainer for ya
Durrrrr my likem chia cause I smrt dev durrrr
This fkn place I sware.
Does anyone know anything anymore?
And another article here
From a forum where people actually know what they are talking about
@jonesjr It seems that you and I are the only ones that are able to put more data on our drives than they are physically able to contain.
You hit 100.3%.
I hit 326.4%
Don’t believe me? Look it up.
Agreed.
When I turned off the harvester io reads logging, I was only hitting 311.2% of disk capacity.
Don’t believe me? Look it up.
Exactly.
I have reported this to both the Chia admins, Microsoft, and when I play my next Doom II multi-player game with Bill Gates, I will bring this to his attention (again), too (he is a skilled Doom II player) – he knows all of the wads.
The last time I talked to Gates about this, he told me he would have it fixed in Windows 11. But I guess that he is pushing it off to Windows 12.
Yes, the harvester does all of that. It writes gigs and gigs of that to the drives that store the plots. That is what is killing your drives.
Don’t believe me? Look it up.
I can’t believe that Gates is pushing this off until Windows 12.
I warned him that he should get this fixed before the Linux kernel developers get the jump on him.
I did not believe you, the last 6 times you wrote that, especially since you never provided evidence that jounaling serves any purpose for storing plots, nor did you provide evidence that when a drive is full that the space that the filing-system allocated specifically for jounaling data, gets crushed by user data.
I did not believe you the last 6 times.
But now that you repeated it 7 or 8 times, it makes you 100% correct.
I will be bringing this up to Gates. Not during our Doom II game (I don’t want to burden him with too many issues). I will bring up the journaling flaw that his nearly trillion dollar company, and his thousands of engineers, have missed for decades. I will bring it up to him when we get together at IHop (he goes for the Chocolate Chip pancakes). His team of personal chefs can’t compete with IHop’s recipe.
No need to continue beating a dead horse.
You beat that horse so hard, that you have been declared victorious.
Beating your dead horse has made you correct.
Providing multiple videos, on general drive performance that has zero to do with storing plots and meeting Chia’s performance guidelines, has also made you correct.
We know that you are correct, because providing multiple, irrelevant videos is more compelling than providing a single video that is relevant.
Tell us more about why journaling matters for storing plots.
3 Likes
Done teaching for the day
Leme know when u sheep are done being so lost
Who cares that they are plots seriously.
It’s meaningless
This goes for all hard drives period
If u seriously take a look at what I’ve posted you will come to understand.
Chia isn’t exempt from the laws of modern computing.
Hard drives are hard drives if they are max full. You will have bad preformance and in my personal experience data loss
End of story
My disks show reads and wright all day long. From my harvesters. It’s faint but it is there.
But that’s only one angle and one point I’m trying to make
Agreed.
Why should we care that they are plots, when it is the plots that are consuming 100% of the drive, and it is the plots that are causing the drive failures, and it is the plots that are killing performance, and it is the plots that are wrecking the journaling.
Why should we care about plots?
I would not have believed you, if you did not add in “period”.
I do understand.
Did you not read that I will be chatting with Bill Gates during our next multi-session Doom II game?
And also when we get together at IHop?
Exactly.
Chia storage is no different than OS storage.
All computer uses are identical.
Don’t believe me? Look it up.
For example, 100% of Google’s servers are identical. Google buys the exact same model servers, regardless of their function.
Don’t believe me? Look it up.
Yes. Exactly.
The performance gets so bad, that I have to wait 10 seconds for notepad to open a 50 byte file.
Don’t believe me? Look it up.
Agreed.
You have repeated the same things enough times to make you correct.
I will add that to my list of topics when Bill and I get together for our knitting session.
He has not shown up for a while, since his divorce (he used to bring his wife).
You have convinced me.
Keep posting more and more articles and videos, regardless of them having anything to do with storing plots. As you wrote “Who cares that they are plots seriously”. With a few more posts, you will convince nearly everyone in this forum.
Tell us more about the importance of journaling, for storing plots.
2 Likes
but the use case just doesn’t matter here… this is just facts.
if its a spinning disk that is near full…
- decreased performance
- no room for io = bad random reads and writes.
- faster fragmentation even if drive isn’t being consistently written to
- no space for log file or cache or any other journaling.
also greatly depends on the filesystem used
things wear down and things break.
a full to the max hard drive is not ideal, and is different for every drive.
why is everyone acting like chia is some sort of magic.
and just doesn’t do what every computer in the world does in regards to caching logging checksumming and auto optimizing…
if you can show me something in GitHub that says that they have mitigated all of this;… im happy to read it…
but ive provided now numerous resources explaining why ANY full disk is a bad thing.
same goes for chia. if u use hard drives.