Uploaded image for project: 'Lustre'
  1. Lustre
  2. LU-14064 socklnd needs improved interface selection and configuration
  3. LU-14107

Keep peer representation consistent across LNet and the LNDs

Details

    • Technical task
    • Resolution: Unresolved
    • Minor
    • None
    • None
    • 9223372036854775807

    Description

      LNet and the LNDs maintain their own representations of peers and these may get out of sync.

      For example, if peer is deleted from LNet (e.g. by running "lnetctl peer del"), the LND still retains the peer and associated connections (easily verified by running "netstat" or the IB equivalent). On the other hand, if peer is brought down and the connections are broken in the LND, LNet still retains the peer.

      There are issues with both of these scenarios. The purpose of this task is to come up with a way to make LNet and the LNDs network views more coherent to help avoid these issues.

       

      Attachments

        Issue Links

          is related to

          Bug - A problem which impairs or prevents the functions of the product. LU-11888 Unreachable client NID confusing Lustre 2.12

          • Minor - Minor loss of function, or other problem where easy workaround is present.
          • Open

          Activity

            [LU-14107] Keep peer representation consistent across LNet and the LNDs
            ashehata Amir Shehata (Inactive) made changes -
            Description Original: LNet and SockLND maintain their own representations of peers and these may get out of sync.

            For example, if peer is deleted from LNet (e.g. by running "lnetctl peer del"), SockLND still retains the peer and associated TCP connections (easily verified by running "netstat"). On the other hand, if peer is brought down and TCP connections are broken in SockLND, LNet still retains the peer.

            There are issues with both of these scenarios. The purpose of this task is to come up with a way to make LNet and SockLND network views more coherent to help avoiding these issues.

                          		New: LNet and the LNDs maintain their own representations of peers and these may get out of sync.

            For example, if peer is deleted from LNet (e.g. by running "lnetctl peer del"), the LND still retains the peer and associated connections (easily verified by running "netstat" or the IB equivalent). On the other hand, if peer is brought down and the connections are broken in the LND, LNet still retains the peer.

            There are issues with both of these scenarios. The purpose of this task is to come up with a way to make LNet and the LNDs network views more coherent to help avoid these issues.

             
            ashehata Amir Shehata (Inactive) made changes -
            Summary Original: Keep peer representation consistent across LNet and SockLND New: Keep peer representation consistent across LNet and the LNDs
            ssmirnov Serguei Smirnov made changes -
            Link New: This issue is related to LU-11888 [ LU-11888 ]
            ssmirnov Serguei Smirnov made changes -
            Assignee Original: WC Triage [ wc-triage ] New: Amir Shehata [ ashehata ]
            ssmirnov Serguei Smirnov created issue -

            People

              ashehata Amir Shehata (Inactive)
              ssmirnov Serguei Smirnov
              Votes:
              0 Vote for this issue
              Watchers:
              1 Start watching this issue

              Dates

                Created:
                Updated: