Killing all peers results in stuck cluster state

[lbradstreet]

Steps to reproduce:

1. Start peers on a particular onyx/id
2. Start a job (possibly unnecessary)
3. Ctrl-c process with peers.
4. Start peers again on same onyx/id

[MichaelDrogalis]

I tested this one. The Aeron media driver is surviving after peers are killed with an embedded media driver. It threw uk.co.real_logic.aeron.driver.exceptions.ActiveDriverException: active driver detected.

[MichaelDrogalis]

The Peer Group inside submit-job collided with the Peer Group on the peers, both running the embedded media driver. Bug confirmed though. The peers come up, but they are never assigned work.

[lbradstreet]

I think I also saw this with core.async messaging.

On 19 Aug 2015, at 4:32 am, Michael Drogalis [email protected] wrote:

The Peer Group inside submit-job collided with the Peer Group on the peers, both running the embedded media driver. Bug confirmed though. The peers come up, but they are never assigned work.

—
Reply to this email directly or view it on GitHub.

[lbradstreet]

I can confirm that this is fixed. Good job.