Uploaded image for project: 'Couchbase PHP client library'
  1. Couchbase PHP client library
  2. PCBC-246

Expose lcb_cntl(..., LCB_CNTL_SKIP_CONFIGURATION_ERRORS_ON_CONNECT, ..)

    Details

    • Type: Task
    • Status: Resolved
    • Priority: Major
    • Resolution: Fixed
    • Affects Version/s: None
    • Fix Version/s: 1.2.0
    • Component/s: None
    • Security Level: Public
    • Labels:
      None

      Description

      There is a special mode which helps users to manage stale bootstrap node list.

      Use case:

      1) setup and deploy application to bootstrap with node1,node2,node3
      2) over the time, admins eject node1 and it might still be listening on 8091
      3) the application should skip LCB_AUTH_ERROR errors and LCB_BUCKET_ENOENT on node1 and continue with node2

      Obviously after successful bootstrap, the list of "known" nodes will be updated for the particular connection. The problem in on-disk configuration

        Issue Links

          Activity

          Hide
          trond Trond Norbye added a comment -

          Should this be the default behavior or not?

          Show
          trond Trond Norbye added a comment - Should this be the default behavior or not?
          Hide
          avsej Sergey Avseyev added a comment -

          I think no. Because we shouldn't add more magic. Right now, the user will get an error and has enough information to figure out why it happened, and to search for possible workarounds. But skipping such nodes by default, might make things more weird, for example if "dead" will became part of new cluster with matching bucket name and password. I know it is possible in current situation, but the user more likely will get errors during transition phase.

          Show
          avsej Sergey Avseyev added a comment - I think no. Because we shouldn't add more magic. Right now, the user will get an error and has enough information to figure out why it happened, and to search for possible workarounds. But skipping such nodes by default, might make things more weird, for example if "dead" will became part of new cluster with matching bucket name and password. I know it is possible in current situation, but the user more likely will get errors during transition phase.
          Show
          trond Trond Norbye added a comment - http://review.couchbase.org/28566

            People

            • Assignee:
              trond Trond Norbye
              Reporter:
              avsej Sergey Avseyev
            • Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved:

                Gerrit Reviews

                There are no open Gerrit changes