Uploaded image for project: 'Core Server'
  1. Core Server
  2. SERVER-29948

Implement stalled I/O testing for storage node watchdog

XMLWordPrintableJSON

    • Type: Icon: Task Task
    • Resolution: Fixed
    • Priority: Icon: Major - P3 Major - P3
    • 3.5.11
    • Affects Version/s: None
    • Component/s: None
    • None
    • Fully Compatible
    • Platforms 2017-07-10, Platforms 2017-07-31

      Using charybdefs, and FUSE on Linux, we can simulate slow/hung I/O in Linux.

      We can specifically trigger a single file or the whole filesystem with calls like this via their Thrift API:

      client.set_fault(['write'], False, 0, 100000, ".*watchdog_probe.txt", False, 50000000, False)

      Recommended Branch with fixes for Direct I/O and a port conflict with Cockpit:
      https://github.com/markbenvenuto/charybdefs/commit/313e0609d4ae917c225151d2c4da554cc9dbaf7c

            Assignee:
            mark.benvenuto@mongodb.com Mark Benvenuto
            Reporter:
            mark.benvenuto@mongodb.com Mark Benvenuto
            Votes:
            0 Vote for this issue
            Watchers:
            6 Start watching this issue

              Created:
              Updated:
              Resolved: