Nodes are not reading hub's configuration during re-registering

[SlavikCA]

Reproduced with Selenium 2.45
HUB: Linux, Ubuntu
Node: Windows 8, Chrome

Steps to reproduce:
1) Start Selenium Grid. Let say, only 2 stations: HUB and NODE.
For HUB, set timeouts, as:
"timeout": 20000
"browserTimeout": 20000
For NODE do not set any timeouts explicitly.
2) Run this code against Grid:

driver.get(url);
sleep(30000);
driver.get(url);

It fails with org.openqa.selenium.remote.SessionNotFoundException.
This line added to the NODE log:
INFO org.openqa.selenium.remote.server.DriverServlet - Session 7f5fffec-4882-4c4c-b091-c780c66d379d deleted due to client timeout
3) So, trying to resolve issue, increase timeouts on the HUB:
"timeout": 40000
"browserTimeout": 40000
Restart HUB. NODE would re-register automatically.

Defect:
Code still fails. Timeout still set at 20 seconds. But, if you look at the Grid console, it would show new (updated) configuration, which is misleading.
The reason: NODE haven't read new timeouts from the HUB.
NODE will only read that new configuration, when re-started.

Expected:
NODE would read configuration from the HUB, when re-registering.

[mach6]

Assigning this to myself for follow up... Will apply any changes (if needed) to Se3.

[mmichaelis]

Observed the same issue when fixing misunderstood timeouts (seconds vs. milliseconds) on hub configuration (docker-selenium). We restarted the hub only but with no result. Restarting nodes helped, though.