Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

deps: update go-plugin to 1.6.2 #24265

Merged
merged 1 commit into from
Oct 21, 2024
Merged

deps: update go-plugin to 1.6.2 #24265

merged 1 commit into from
Oct 21, 2024

Conversation

tgross
Copy link
Member

@tgross tgross commented Oct 21, 2024

During allocation recovery, the go-plugin reattachment workflow checks to see if the PID we stored in client state is still running. If so, we try to connect to that process. If that fails, we kill the process under the presumption it's not working. But during reattachment we don't know that the PID we have is still valid. Which means that the process we're trying to attach to may have exited and a different process has spawned with the same PID. This results in some unrelated process getting silently killed.

Update go-plugin to 1.6.2, which includes a bug fix so we do not kill the process.

Fixes: #23969
Ref: https://hashicorp.atlassian.net/browse/NET-11233

@tgross tgross added this to the 1.9.2 milestone Oct 21, 2024
@tgross tgross added theme/dependencies Pull requests that update a dependency file type/bug labels Oct 21, 2024
@tgross tgross requested review from shoenig and schmichael October 21, 2024 18:42
@tgross tgross marked this pull request as ready for review October 21, 2024 18:42
@tgross tgross added backport/ent/1.7.x+ent Changes are backported to 1.7.x+ent backport/ent/1.8.x+ent Changes are backported to 1.8.x+ent backport/1.9.x backport to 1.9.x release line labels Oct 21, 2024
@tgross
Copy link
Member Author

tgross commented Oct 21, 2024

Build broken by hashicorp/go-plugin#312. Need to fix that up. (Follow-up: https://hashicorp.atlassian.net/browse/NET-11447)

During allocation recovery, the go-plugin reattachment workflow checks to see if
the PID we stored in client state is still running. If so, we try to connect to
that process. If that fails, we kill the process under the presumption it's not
working. But during reattachment we don't know that the PID we have is still
valid. Which means that the process we're trying to attach to may have exited
and a different process has spawned with the same PID. This results in some
unrelated process getting silently killed.

Update go-plugin to 1.6.2, which includes a bug fix so we do not kill the
process.

Fixes: #23969
@tgross tgross merged commit 6d17a9b into main Oct 21, 2024
26 checks passed
@tgross tgross deleted the go-plugin-1.6.2 branch October 21, 2024 19:48
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
backport/ent/1.7.x+ent Changes are backported to 1.7.x+ent backport/ent/1.8.x+ent Changes are backported to 1.8.x+ent backport/1.9.x backport to 1.9.x release line theme/dependencies Pull requests that update a dependency file type/bug
Projects
None yet
Development

Successfully merging this pull request may close these issues.

process not managed by Nomad killed on restart
2 participants