Uploaded image for project: 'Couchbase Server'
  1. Couchbase Server
  2. MB-19650

ryow throughoutput drops significantly 4.5.0-2514 vs 4.5.0-2548

    XMLWordPrintable

Details

    • Bug
    • Status: Closed
    • Critical
    • Resolution: Won't Fix
    • 4.5.0
    • 4.5.0
    • performance
    • Untriaged
    • Centos 64-bit
    • Unknown

    Attachments

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

      Activity

        Hi, John,

        Can you please help us take a look at the logs?

        Thanks,
        Qi

        qi Qi Zhu (Inactive) added a comment - Hi, John, Can you please help us take a look at the logs? Thanks, Qi
        jliang John Liang added a comment - - edited

        In 2514, I see incoming mutation rate into indexer is about 2.1K/s. In 2548, I see mutation rate is about 373/s. Wait time increase correspondingly between 2514 and 2548 (from 6ms to 2s)

        jliang John Liang added a comment - - edited In 2514, I see incoming mutation rate into indexer is about 2.1K/s. In 2548, I see mutation rate is about 373/s. Wait time increase correspondingly between 2514 and 2548 (from 6ms to 2s)
        jliang John Liang added a comment -

        From 2584, node .73, mutation rate is about 85/s per node over a 10-min interval. That's roughly ~340/s per cluster/4-node.

        2016-05-16T08:10:35.163-07:00,ns_1@172.23.100.73:
        cmd_set 5063308

        2016-05-16T08:20:35.163-07:00,ns_1@172.23.100.73:
        cmd_set 5114816

        For 2514, node .73, mutation rate is 552/s per node over 10-min interval (~2K/s per cluster)

        2016-05-16T07:10:40.478-07:00,ns_1@172.23.100.73:
        cmd_total_sets 5354480

        2016-05-16T07:00:40.477-07:00,ns_1@172.23.100.73:
        md_total_sets 5022769

        jliang John Liang added a comment - From 2584, node .73, mutation rate is about 85/s per node over a 10-min interval. That's roughly ~340/s per cluster/4-node. 2016-05-16T08:10:35.163-07:00,ns_1@172.23.100.73: cmd_set 5063308 2016-05-16T08:20:35.163-07:00,ns_1@172.23.100.73: cmd_set 5114816 For 2514, node .73, mutation rate is 552/s per node over 10-min interval (~2K/s per cluster) 2016-05-16T07:10:40.478-07:00,ns_1@172.23.100.73: cmd_total_sets 5354480 2016-05-16T07:00:40.477-07:00,ns_1@172.23.100.73: md_total_sets 5022769
        jliang John Liang added a comment -

        We need to find out why the mutation rate is different between the 2 runs.

        jliang John Liang added a comment - We need to find out why the mutation rate is different between the 2 runs.

        I have noticed a n1ql throughput drop in build 2548 in CI2, CI3, Q2,Q3, stale=false test.

        Sent out an email to Wayne to confirm.

        qi Qi Zhu (Inactive) added a comment - I have noticed a n1ql throughput drop in build 2548 in CI2, CI3, Q2,Q3, stale=false test. Sent out an email to Wayne to confirm.

        2548 is not a good build. There's a DCP regression which caused many tests to fail.

        Close this bug as won't fix.

        Waiting for new build.

        qi Qi Zhu (Inactive) added a comment - 2548 is not a good build. There's a DCP regression which caused many tests to fail. Close this bug as won't fix. Waiting for new build.

        People

          andreibaranouski Andrei Baranouski
          andreibaranouski Andrei Baranouski
          Votes:
          0 Vote for this issue
          Watchers:
          4 Start watching this issue

          Dates

            Created:
            Updated:
            Resolved:

            Gerrit Reviews

              There are no open Gerrit changes

              PagerDuty