Dynamic queueing of jobs between proxies

[dellock6]

Hi all, I've done some tests on our infrastructure and I would like to share them with you all, hoping for some other experience or ideas.

First, a quick description of the lab: 3 backup jobs (J1, J2 and J3), 10VMs each, and 2 Veeam proxies (call them P1 and P2). All jobs start at the same time at 21:00, and cuncurrency limit is set to 2.
Correctly, J1 and J2 starts each on every proxy, and J3 gets in the queue waiting.

What I saw, and was strange to me, is I had to stop J2 for some reasons, but J3 was not starting anyway, even if P2 had no job running. While J1 was still running on P1, I had to manually stop J3, and then restart it. In this second run, it correctly found P2 free and used it.

Seems like the queued job had chosen the proxy before even starting; but for what I always knew, job queue was dynamic, isn't it?

Luca.

[Gostev]

Seems like the queued job had chosen the proxy before even starting; but for what I always knew, job queue was dynamic, isn't it?

Correct, it is dynamic - proxies are selected in real-time. How long did you wait after stopping J2?

[dellock6]

Uhm, not so much to be honest. Do you know the timeout before a new proxy is picked? Just to be sure it was only me beeing to hurried.

Luca.

[tsightler]

I think the default is about 60 seconds when a job is waiting on resources.

[dellock6]

Uhm, ok so it could be I simply did not wait enough, thanks for clarification.

Luca.