#1132852 monerod throws Exception: std::runtime_error

Package:
librandomx0
Source:
librandomx0
Description:
proof of work (PoW) algorithm for CPUs
Submitter:
Emma Müller
Date:
2026-06-05 01:23:01 UTC
Severity:
normal
#1132852#5
Date:
2026-04-06 11:22:08 UTC
From:
To:
I decided to store the blockchain on a NTFS drive with enough space. Here is the mount line in /etc/fstab:

UUID=…   /media/my_NTFS_Drive    ntfs3   rw,relatime,discard,windows_names,gid=users,fmask=0000,dmask=0000       0       0

Shortly before the initial sync was over, around 98%, I started getting errors in the log:

2026-04-06 10:27:05.533     7f5a166fa6c0        INFO    stacktrace      src/common/stack_trace.cpp:134  Exception: std::runtime_error

followed by a stacktrace. This is informative but unactionable. systemctl --user restart monerod didn't help. The log is attached.

$ cat ~/.config/systemd/user/monerod.service
[Unit]
Description=Monero Node
After=network.target

[Service]
Type=simple
ExecStart=/usr/bin/monerod \
   --data-dir /media/my_NTFS_Drive/path_to_my_monero_dir \
   --prune-blockchain \
   --log-file /media/my_NTFS_Drive/path_to_my_monero_dir/bitmonero.log \
   --non-interactive
Nice=19
IOSchedulingClass=idle
Restart=always
RestartSec=30

[Install]
WantedBy=default.target



Please fix this. (Guesses: packaging a newer version might be due, or some debugging option has to be turned on first, or the bug report has to be reassigned to the ntfs3 maintainers, ….)
#1132852#10
Date:
2026-04-06 12:07:43 UTC
From:
To:
Correction:

I started to get INFO messages (and not errors). Still, they are clutter.

Now I appended

Environment=MONERO_RANDOMX_UMASK=1

to the [Service] section, and these INFO messages with exceptions and stacktraces seem to be gone for now. Not sure whether this is a permanent fix or whether they will reappear and more has to be done.

 From a programming point of view on the original setup, a constantly thrown exception is wasteful; detecting the file system and adapting to its capabilities properly right from the startup of monerod would be a better way.
#1132852#15
Date:
2026-06-04 17:39:13 UTC
From:
To:
This is fixed in librandomx v2.0.1 which we cannot use with current monero.
So this needs backporting.
#1132852#24
Date:
2026-06-04 20:58:06 UTC
From:
To:
Sorry for the noise. Got the wrong issue. For this one:

Fix: Enable hugepages on the host:
sudo sysctl -w vm.nr_hugepages=1280
# Make persistent:
echo "vm.nr_hugepages=1280" | sudo tee -a /etc/sysctl.d/monero.conf

Or launch monerod without --randomx-large-pages if you explicitly set that flag.

Please note: NTFS under Linux has no mlock support, which can exacerbate the large-page issue.
#1132852#33
Date:
2026-06-05 01:21:00 UTC
From:
To:
If I do this, should I keep or drop

Environment=MONERO_RANDOMX_UMASK=1

in the [Service] section of ~/.config/systemd/user/monerod.service ?

I don't think I use --randomx-large-pages.

Gratefully!
Emma