Menu

#3008 fmd: improve failover time

5.19.03
fixed
Gary Lee
None
defect
unknown
-
major
False
2019-02-19
2019-02-18
Gary Lee
No

Currently, if split brain prevention is enabled, fmd waits a few seconds to process MDS topology messages before performing a failover. There is no need to do this if we don't prioritise network partition size.

related ticket for rde [#3005]

Related

Tickets: #3005
Wiki: ChangeLog-5.19.03

Discussion

  • Gary Lee

    Gary Lee - 2019-02-18
    • Description has changed:

    Diff:

    --- old
    +++ new
    @@ -1 +1,3 @@
     Currently, if split brain prevention is enabled, fmd waits a few seconds to process MDS topology messages before performing a failover. There is no need to do this if we don't prioritise network partition size.
    +
    +related ticket for rde [#3005]
    
     

    Related

    Tickets: #3005

  • Gary Lee

    Gary Lee - 2019-02-19
    • status: accepted --> fixed
     
  • Gary Lee

    Gary Lee - 2019-02-19

    commit 8ccffc2cd9cd117578227e9cd49421e5c578fec6
    Author: Gary Lee gary.lee@dektech.com.au
    Date: Tue Feb 19 14:57:53 2019 +1100

    rded: do not send SUCCESS to main thread [#3008]
    
    do not send RDE_MSG_ACTIVE_PROMOTION_SUCCESS to
    main thread if lock cannot be obtained
    

    commit 28e17d107f4a079155e03d9f875a3c0262ea19f5
    Author: Gary Lee gary.lee@dektech.com.au
    Date: Tue Feb 19 14:57:53 2019 +1100

    fmd: improve failover response time [#3008]
    
    Improve failover response time if split brain prevention is enabled
    but FMS_TAKEOVER_PRIORITISE_PARTITION_SIZE is set to 0.
    
    Also, return immediately if node promotion fails to avoid
    sending active role to RDA.
    
     

Log in to post a comment.

MongoDB Logo MongoDB