Skip to content

Ipfs no longer shuts down properly #682

Closed

Description

@whyrusleeping

I dont know if this has been filed already (if so, just close this one) but killing ipfs with SIGINT no longer succeeds. This should be alpha priority.

Activity

  1. added this to the α milestone on Jan 28, 2015
  2. btc commented on Jan 28, 2015

    @btc
    Contributor

    I've found it helpful to include a Debug log statement in Close() error methods of large components. Makes it easy to see when things shut down as they should.

  3. whyrusleeping commented on Jan 29, 2015

    @whyrusleeping
    MemberAuthor

    It appears that dialAddrs in the swarm doesnt respect the context as well as it should, and even farther down, rawConnDial doesnt really use the context, but just sets a timeout. This might be part of the issue, im still investigating.

  4. whyrusleeping commented on Jan 29, 2015

    @whyrusleeping
    MemberAuthor

    looks like i noticed this before in #650

  5. whyrusleeping commented on Jan 29, 2015

    @whyrusleeping
    MemberAuthor

    I beleive #690 is related as well.

  6. whyrusleeping commented on Jan 29, 2015

    @whyrusleeping
    MemberAuthor

    More experiment notes: If i leave my daemon on for a while, and disconnect my internet connection, the daemon appears to close down properly.

  7. jbenet commented on Jan 29, 2015

    @jbenet
    Member

    i've observed the same. it's related to network connections. probably
    contexts not cancelling properly.

    On Thu, Jan 29, 2015 at 2:54 PM, Jeromy Johnson notifications@github.com
    wrote:

    More experiment notes: If i leave my daemon on for a while, and disconnect
    my internet connection, the daemon appears to close down properly.

    —
    Reply to this email directly or view it on GitHub
    #682 (comment).

  8. whyrusleeping commented on Feb 3, 2015

    @whyrusleeping
    MemberAuthor

    The next time anyone encounters the daemon not quitting, could you wait up to one minute? Im thinking that it might be waiting for dial calls to timeout, if thats the case, the daemon should exit once the timeout triggers and we know where to direct our debugging efforts

  9. whyrusleeping commented on Feb 3, 2015

    @whyrusleeping
    MemberAuthor

    Just confirmed that the daemon will shutdown at some point after the kill signal is sent, < 30s in my cases.

  10. btc commented on Feb 3, 2015

    @btc
    Contributor

    I found it useful to add log statements in the Close() methods of sub-systems to see the order in which they shutdown.

  11. added a commit that references this issue on Oct 23, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Labels

kind/bugA bug in existing code (including security flaws)

Type

No type

Projects

No projects

    Milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions