Add OSRM_LOCK_DIR environment variable for containerized deployments (memory leak fix) #7312
+28
−5
Conversation
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
Allows specifying a custom directory for shared memory lock files via the OSRM_LOCK_DIR environment variable. This enables containerized deployments to persist lock files across container restarts, preventing crashes when using shared memory (osrm-routed -s)
wes4m
changed the title
TheMarex
reviewed
Jan 7, 2026
Member
TheMarex
left a comment
TheMarex
left a comment
There was a problem hiding this comment.
This change makes sense. Please add an entry to the changelog. Let me know if you have time to do the small edits yourself, otherwise I can try to put this on my TODO for the week.
include/storage/shared_memory.hpp
Outdated
| std::filesystem::path dir(lock_dir); | ||
| if (!std::filesystem::exists(dir)) | ||
| { | ||
| std::filesystem::create_directories(dir); |
Member
There was a problem hiding this comment.
IMHO this should error here. The directory should exist.
include/storage/shared_memory.hpp
Outdated
| // Returns directory for OSRM lock files (OSRM_LOCK_DIR env var or system temp) | ||
| inline std::filesystem::path getLockDir() | ||
| { | ||
| if (const char *lock_dir = std::getenv("OSRM_LOCK_DIR")) |
Member
There was a problem hiding this comment.
Generally speaking I think a prefix like OSRM_ makes sense, but it is inconsistent with the other env variables. I would recommend SHM_LOCK_DIR.
wes4m
force-pushed
the
fix/container-shm-lock
branch
from
019fbb9 to
26eb039
Compare
Author
Thanks for the review @TheMarex. changes pushed. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.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.
Problem
OSRM crashes with
"lock file does not exist"when using shared memory (osrm-routed -s) in containerized environments (Kubernetes) after a container restart.The crash triggers a restart/crash loop, and each restart accumulates orphaned shm segments, leaking memory proportional to the graph size until the node is exhausted.
Related issues: #5134, #5703
Reproduction
See gist: https://gist.github.com/wes4m/719cb69b72e26c09c7ff57ed71cf33d9
Why This Happens in Containers
OSRM's shared memory implementation assumes
/tmpand System V shared memory have the same lifecycle. This is true on traditional systems where both are cleared on reboot, but not in containers:/tmpfilesystemWhen a container restarts:
/tmpis reset and lock files get deletedSharedRegionRegister(in shm) still references old segmentsSolution
This PR adds
OSRM_LOCK_DIRenvironment variable to specify a custom directory for lock files. This allows containerized deployments to place lock files in a volume that persists across container restarts (e.g. kubernetesemptyDir)Other Approaches
Mounting
/tmpto a persistent volume fixes the issue but will persist all temporary files from the container, not just lock files. SettingTMPDIRwill do the same, and cleaning orphaned shm segments on startup is a workaround that doesn't address the root cause.OSRM_LOCK_DIRonly affects lock file location and is backward compatible (tmp dir fallback), behavior is unchanged.