Repository navigation
fix: raise the container memlock limit so QEMU can set up io_uring - #9
Merged
Merged
Conversation
QEMU 10.1+ uses io_uring for its event loop and, on recent kernels, for the disk. The rings are locked memory and the container default is a few hundred KB, so QEMU exited at boot with "Failed to initialize io_uring: Cannot allocate memory". Unlimited memlock fixes both uses at once. DISK_IO is removed again: 2.0.1 and 2.0.2 pointed at the disk backend, which was not the cause.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Current Behaviour
On some hosts the first start ends right after "Booting Windows using QEMU" with
and restarts in a loop. 2.0.1 and 2.0.2 attributed this to the disk I/O backend (
DISK_IO), but QEMU 10.1+ also uses io_uring for its main event loop, soDISK_IO=nativedid not help. The real limit is the container's locked memory (memlock), a few hundred KB by default, which io_uring's rings exceed.Expected Behaviour
docker-compose.ymlsetsulimits.memlockto unlimited for the service, the same setting QEMU and Elasticsearch documentation recommend. It is not a privilege and needs no capability.DISK_IOis removed from the compose file,.env.exampleand the README; QEMU picks the backend.Confirmed on the affected host (Ubuntu 20.04, kernel 5.4): with the memlock limit raised QEMU boots and the Windows install proceeds.
docker compose configpasses.