10-27-2015 04:08 PM
I've checked the NIOS admin guide and don't see it mentioned...
I want to block MAC address from our network.
I've added the classes and MAC address and now it asks me to restart services on both appliances.
We do not use HA. We have a GRID of two appliances.
Will this interrupt our users? Anything I can do so that it doesn't?
Thanks for any advice.
Solved! Go to Solution.
10-28-2015 12:22 PM
For anyone else that is looking for an answer...
I sent a support ticket asking this and they replied that it does indeed interrupt DHCP services but only for a few moments and of course any host with a DHCP address already wouldn't notice a thing.
10-28-2015 12:44 PM
The "restart services" function stops and then restarts DNS and/or DHCP services in order to load new configuration settings. This typically does not take very long. I just tested this on my three-member grid (two DNS/DHCP members plus a reporting server) by making a change to DHCP grid properties and then restarting. Since the change was made to DHCP grid properties it required restarting DHCP (but not DNS) services on both grid members; by default this is done sequentially, one immediately after the other. Per the logs on the affected grid members, the restart took less than 10 seconds per grid member, with the whole restart sequence being completed in less than 30 seconds. (It would of course take longer if you had more grid members.)
As for the impact to users and how you can minimize it: During the period while a service is restarting on a grid member that service will be unavailable to users. For example, in my case when restarting DHCP services there was a sub-10-second period when each grid member was not able to service DHCP lease requests. Similar considerations would apply to restarting DNS services; there would be a very brief period during which DNS queries were not serviced.
Both the DHCP and DNS protocols allow for retries of requests, so in practice the actual impact to users should be minimal to non-existent. This is especially true for DHCP---with long lease times it's quite possible that there may no incoming lease requests at all during the restart period.
You can also make sure that you have some level of service redundancy between your two grid members. For example, for a given DNS zone you could specify both grid members as nameservers, for DHCP you could tell clients that both grid members should be used a recursive DNS servers, etc.
There are fancier things you can do as well: You can schedule restarts for off-hours, you can (in 7.2 at least) define "restart groups" of grid members that get their services restarted at a different time than other restart groups, and so on. However for a relatively small deployment that may not be worth it.