Changes between Initial Version and Version 1 of Ticket #27066, comment 7


Ignore:
Timestamp:
Aug 9, 2018, 7:41:52 PM (14 months ago)
Author:
mikeperry
Comment:

Legend:

Unmodified
Added
Removed
Modified
  • Ticket #27066, comment 7

    initial v1  
    11Ok based on the above, my suspicions are accumulating around some kind of interaction between timed out circuits and the circuit failure limit, which would explain why your services stop working. I do not know where this CBT error is coming from though. #27049 is one way that it can happen, but if there are no loglines related to that, then something else must be happening.
    22
    3 I also do not know why this limit is only affecting v3 services. Perhaps there is some v3-specific code that counts failures differently than v2.
     3I also do not know why this limit is only happening when v3 services are enabled. Perhaps there is some v3-specific code that counts failures differently than v2.
    44
    55Are you able to run branches with additional logging and/or potential fixes? I have been unable to reproduce this on a v3-only service that is under continual use via synthetic wget activity.