ATS 10.0.4 intermittent SIGSEGV in FetchSM/PluginVC callback path (2 crashes, same stack pattern)
- Dominant language
- C++
- Stars
- 2k
- Forks
- 874
- Avg merge
- 6d 15h
- Merged PRs (30d)
- 46
Description
Summary
We observed two ATS crashes on version 10.0.4 with the same failure pattern.
Both crashes are SIGSEGV and both fail in the same stack path.
Environment
Apache Traffic Server 10.0.4
Linux x86_64, kernel 4.18.0-553.109.1.el8_10.x86_64
Crash events
2026-04-16 22:10:54 UTC
2026-04-16 22:21:22 UTC
Systemd (both hosts)
Main process exited, code=killed, status=11/SEGV
Service restarted by systemd
Common stack signature (both crash logs)
crash_logger_invoke(int, siginfo_t*, void*) + 0x72
FetchSM::InvokePluginExt(int) + 0x256
FetchSM::process_fetch_read(int) + 0x9f
FetchSM::fetch_handler(int, void*) + 0xd8
PluginVC::process_write_side() + 0x96d
PluginVC::main_handler(int, void*) + 0x366
EThread::process_event / process_queue / execute_regular
Key observation
The same function and same offset appear in both crashes:
FetchSM::InvokePluginExt + 0x256
This suggests a repeatable fault in the same execution path.
What we are NOT claiming
We are not attributing this to a specific traffic profile.
We are not attributing this to a specific plugin/remap.
We are not claiming OOM or system memory pressure as root cause.
Request
Is this a known issue in ATS 10.0.4?
Is there a fix in a newer release (10.0.x or 10.1.x)?
Are there recommended mitigations/workarounds for this stack path?
Available
Full crash logs for both events
Exact systemd/journal excerpts around both crash timestamps
[edge-2.txt](https://github.com/user-attachments/files/26821455/edge-2.txt)
[edge-1.txt](https://github.com/user-attachments/files/26821453/edge-1.txt)
[crash.log.gz](https://github.com/user-attachments/files/26821456/crash.log.gz)
[crash-2026-04-16-221054.log.gz](https://github.com/user-attachments/files/26821454/crash-2026-04-16-221054.log.gz)
Contributor guide
Assessment
This issue has not been assessed yet.