Uploaded image for project: 'Couchbase Monitoring and Observability Stack'
  1. Couchbase Monitoring and Observability Stack
  2. CMOS-154

Add check for DCP names > 200 chars (MB-34280)

    XMLWordPrintable

Details

    Description

      This has hit quite a few people and could cause serious issues when upgrading from 6.x to 7. Be good to have a check for it.

      DCP names are exposed when requesting the stats group dcp over the memcached protocol - we already have code that uses this data, should be possible to piggyback on it.

      Attachments

        Issue Links

          No reviews matched the request. Check your Options in the drop-down menu of this sections header.

          Activity

            marks.polakovs Marks Polakovs (Inactive) created issue -
            marks.polakovs Marks Polakovs (Inactive) made changes -
            Field Original Value New Value
            Rank Ranked higher
            marks.polakovs Marks Polakovs (Inactive) made changes -
            Status To Do [ 10001 ] In Progress [ 3 ]
            marks.polakovs Marks Polakovs (Inactive) made changes -
            Rank Ranked lower
            marks.polakovs Marks Polakovs (Inactive) made changes -
            Link This issue relates to CBSE-11034 [ CBSE-11034 ]
            marks.polakovs Marks Polakovs (Inactive) made changes -
            Link This issue relates to MB-34280 [ MB-34280 ]
            marks.polakovs Marks Polakovs (Inactive) made changes -
            Status In Progress [ 3 ] In Review [ 10107 ]
            marks.polakovs Marks Polakovs (Inactive) made changes -
            Status In Review [ 10107 ] In Progress [ 3 ]

            Moving back to In Progress as I noticed a mistake in the patch (right after merging it...) - the tests have stream names in the format "eq_dcpq:replication:ns_1@127.0.0.1->ns_2@127.0.0.1:bucket:max_buffer_bytes" but they shouldn't include the stat name at the end (and since the tests pass, that implies the code itself is wrong).

            marks.polakovs Marks Polakovs (Inactive) added a comment - Moving back to In Progress as I noticed a mistake in the patch (right after merging it...) - the tests have stream names in the format "eq_dcpq:replication:ns_1@127.0.0.1->ns_2@127.0.0.1:bucket:max_buffer_bytes" but they shouldn't include the stat name at the end (and since the tests pass, that implies the code itself is wrong).

            Build couchbase-cluster-monitor-1.0.0-129 contains cbmultimanager commit 8e28b25 with commit message:
            CMOS-154 Add checker for long DCP names

            build-team Couchbase Build Team added a comment - Build couchbase-cluster-monitor-1.0.0-129 contains cbmultimanager commit 8e28b25 with commit message: CMOS-154 Add checker for long DCP names
            marks.polakovs Marks Polakovs (Inactive) made changes -
            Status In Progress [ 3 ] In Review [ 10107 ]
            Ilias.kourtoudis Ilias Kourtoudis made changes -
            Labels Support_priority
            marks.polakovs Marks Polakovs (Inactive) made changes -
            Epic Link CMOS-32 [ 171120 ]

            Verifying this is tricky, because I'm struggling to create a cluster with hostnames such that the DCP stream names are longer than 200 characters but shorter than (255 - length of the longest DCP stat name), otherwise memcached will fail to send the stats reply we're looking for and thus the checker will fail.

            Now that we can do Loki alerts (cf. CMOS-211) I might add a couple rules there for detecting the telltale log messages, that way we can catch this even if Cluster Monitor can't.

            marks.polakovs Marks Polakovs (Inactive) added a comment - Verifying this is tricky, because I'm struggling to create a cluster with hostnames such that the DCP stream names are longer than 200 characters but shorter than (255 - length of the longest DCP stat name), otherwise memcached will fail to send the stats reply we're looking for and thus the checker will fail. Now that we can do Loki alerts (cf. CMOS-211 ) I might add a couple rules there for detecting the telltale log messages, that way we can catch this even if Cluster Monitor can't.
            pvarley Patrick Varley made changes -
            Status In Review [ 10107 ] In Progress [ 3 ]
            pvarley Patrick Varley made changes -
            Resolution Done [ 6 ]
            Status In Progress [ 3 ] Done [ 10002 ]

            People

              marks.polakovs Marks Polakovs (Inactive)
              marks.polakovs Marks Polakovs (Inactive)
              Votes:
              0 Vote for this issue
              Watchers:
              3 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved:

                Gerrit Reviews

                  There are no open Gerrit changes

                  PagerDuty