MR Batch Processor panicked at MmapPageApplicator failed to mmap bytes to memory file

Developers
1

On my phone and using photos to report this error. We were at an in-person hackathon this past weekend and we found this problem (or two very similar problems) when using the replica with dfx 0.15.0.

Either on dfx deploy or on replica startup, it’s hard to tell or remember, these errors would be reported. It would then be impossible to actually finish a dfx deploy:

thread 'MR Batch Processor' panicked at 'MmapPageAllocator failed to mmap 765952 bytes to memory file #33 at offset 0 for deserialization: ENOEXEC: Exec format error'

And then I believe the true blockers on deploy were these:

tcp connect error, Connection Refused

Machine specs I could gather:

AMD Ryzen 5 3500U with Radeon Vega Mobile Gfx 2.10 Ghz, 8 GiB of RAM, Matebook D 15, Windows 10 Home 22H2, using WSL

This does seem similar to Errors creating a canister with 'dfx deploy'

@roman-kashitsyn

1 Like
2

The error comes from here: mmap.rs - dfinity/ic - Sourcegraph

I think it’s related to WSL quirks; I’m sure @ulan can shed more light on this issue.

3 Likes
3

hi Jordan, that does indeed look like a WSL issue. Are you by any chance using WSL 1? Our recommendation is to use WSL 2 (see here: Installing the IC SDK | Internet Computer).

1 Like
4

These weren’t issues with my machine so I don’t know. I do know that these users had just installed WSL (at least one of them), thus I don’t know if that means they had the chance to install WSL 1 or not.

1 Like
5

Sorry for the late reply, I was travelling. +1 to the theory about WSL 1 vs WSL 2.

6

I could confirm this theory: I could reproduce the issue on WSLv1, but not on WSLv2.

1 Like