Details

    • Type: New Feature
    • Status: Resolved
    • Priority: Minor
    • Resolution: Duplicate
    • Fix Version/s: None
    • Component/s: None
    • Labels:
      None

      Description

      We need a slow query log for Cassandra similar to the one in mysql.
      Tracing does not work because you don't want to enable it for all the requests and want to catch the slow queries.
      We already have a JIRA to display queries which are going over lot of tombstones. But a query can be slow also because it is returning lot of data.
      We can store all the slow queries for a day in system table.

        Issue Links

          Activity

          Hide
          jbellis Jonathan Ellis added a comment -

          Tracing works fine at 0.1% or similar.

          Show
          jbellis Jonathan Ellis added a comment - Tracing works fine at 0.1% or similar.
          Hide
          rustyrazorblade Jon Haddad added a comment - - edited

          I agree that a slow query log would be very useful. What I personally need is a log where I can say "slow_query_time = 3s". Tracing just generates a lot of noise that I have to wade through. Additionally, tracing can only identify a general query pattern that is being non-performant, rather than an outlier (a partition with a ton of tombstones).

          On the mysql side, I've used tools like mysqldumpslow to aggregate slow queries and do analysis of my system. From an operational perspective this would help give feedback to developers.

          Show
          rustyrazorblade Jon Haddad added a comment - - edited I agree that a slow query log would be very useful. What I personally need is a log where I can say "slow_query_time = 3s". Tracing just generates a lot of noise that I have to wade through. Additionally, tracing can only identify a general query pattern that is being non-performant, rather than an outlier (a partition with a ton of tombstones). On the mysql side, I've used tools like mysqldumpslow to aggregate slow queries and do analysis of my system. From an operational perspective this would help give feedback to developers.
          Hide
          kohlisankalp sankalp kohli added a comment -

          The dupe JIRA talks about slow queries which have timed out. Queries can still be slow but not timeout. The other JIRA is actually timeout query log.

          Show
          kohlisankalp sankalp kohli added a comment - The dupe JIRA talks about slow queries which have timed out. Queries can still be slow but not timeout. The other JIRA is actually timeout query log.
          Hide
          rustyrazorblade Jon Haddad added a comment -
          Show
          rustyrazorblade Jon Haddad added a comment - +1 sankalp kohli

            People

            • Assignee:
              Unassigned
              Reporter:
              kohlisankalp sankalp kohli
            • Votes:
              3 Vote for this issue
              Watchers:
              7 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved:

                Development