Learn More
Resend Email
5
klekih
7y

So I'm working on some communications app which bridges the main server/database with some equipment on the field. Now this equipment works in a redundancy pair: two cpus, A and B, both connected via ethernet, one active, the other standby/replicating and no comms on it. One obvious requirement is that when this equipment swaps the active cpus the comms app should switch as well. Fair enough, going with this into testing phase.

This guy, from qa, got some instructions from someone else:
1. Trigger the soft switch from equipment so that cpus got swapped.
2. Remove the ethernet cable from the standby unit.
3. Observe the communications.

And the test goes like: cpu A is active, B is standby. Switch is triggered, B becomes active, A goes into standby. The cable is pulled out from cpu B.

Test result: failed, no comms ever

undefined

testing

story
Favorite
Ranter
Comments
  • 1
    7y
    If B has no comms should it switch back to A? Otherwise this test shows an architecture flaw.
  • 0
    7y
    @bkwilliams during manual switching, no, it should stay where you want. Otherwise, you're right.
  • 0
    7y
    @tomb if the hardware is out somewhere in a pelican case you probably want to be able to do a "wake on LAN" thing to override a manual setting. If it's a jumper setting and you have a bad cable/port, you wouldn't be down with redundant hardware.
  • 1
    7y
    @bkwilliams I don't have any control over how the hardware works down there in the field. So this manual switch is intended to simulate when a cpu goes off completely. Hence the cable pulled. Otherwise there are auto-procedures to recover it.
    Should have been more clear on the story, maybe.
Related Rants
devRant © 2021 Hexical Labs LLC
Privacy Policy  |  Terms of Service
Add Comment