Plotting in WSL2 consumes more space than expected in temp drive.

Hi, I followed the installation instruction for WSL2. Increased the maximum capacity of WSL2 to 2TB(Disk /dev/sdb: 1.98 TiB, 2147483648000 bytes, 4194304000 sectors) and started 2 processes to run chia plots create -k 32 -b 6750 -t /mnt/p -d /mnt/d -n 1 -r 4. The temp drive(/mnt/p) has 1.8TB available space and was empty before plotting. Now I can see that 865GB are consumed when both processes are in phrase 1, computing table 5 (complete log files attached below). According to the wiki, with k=32 plot the temp space used is 256.6GB, and this matches the behavior when I plot in Windows with GUI. So I have no idea why more than 2*256.6 GB is used with 2 plots in WSL2. Confirmed that there is no other plotting process running.

For debugging purposes, I stopped plotting processes and move temp files to the trash. I still see that there is 771GB space used, although I cannot see any file in it through windows, with hidden files visible. I have to format the disk to make it empty again.

I see a similar issue happened in another windows PC with WSL2 plotting so I think it’s not an intermittent issue. I suspect that some of my configurations are wrong or Windows couldn’t identify the space correctly when the WSL2 is operating on it. Did anyone else encounter similar issue?

logfile-1.txt logfile-2.txt

[zeroarst]

I have exactly the same issue. I have 1TB NVMe SSD ram. I ran 3 plots in parallel. Then all ate up all space. I used GUI with the same approach dose not give me the problem.

[zrjaa1]

Reproduced the issue by creating 5 plot processes chia plots create -k 32 -b 6750 -t /mnt/p -d /mnt/d -n 1 -r 4. The 1.83 temp drive gets full really quick and I see some logs similar to Only wrote 0 of 340764 bytes at offset 7900263522 to "/mnt/p/plot-k32-2021-04-24-16-47-872e01a27f1c03db0ac531020ac3defaa8ff40b31c60992c9c397a58000e941c.plot.table4.tmp"with length 7900266496. Error 1. Retrying in five minutes. With an disk analysis utility called wiztree I get something like this:

Please note that although it shows only 9.8GB occupied by P:\$Extend\$Deleted at the moment of the screenshot, when I first check the analysis it says 1TB in total. According to The “$Extend$Deleted” directory, this indicates that a delete operation on those files has been performed but at least one process is still using them at the time so they’re actually deleted later. Apprantly it goes away fast(about 10 minutes after the disk is full), but I’m wondering if it’s expected behavior of plotting because 5 plots actually occupy more(at least 1.83TB used) than 5 * 256.6GB = 1.283TB of temp drive space.

Edit: It’s not a one-time thing during the plotting process. I’ve noticed that another 180GB shows up in the P:\$Extend\$Deleted close to the end of phase 1. So I think it’s more like a periodic issue.