Wonder if anyone can help, I feel I might have to log a support call.
We have 3 Web Gateway Boxes all on version 6.8.7 build 7306 and 1 is set to Master and 2 to Site instance.
The 2 site instances lost commnication with the master and all 3 boxes needed rebooting to fix the issue.
The installation was carried out by an external company and I was wondering how the cluster actually works?
The documentation states that the Master is just for Administration purposes but the incident that occured for us suggest its a little bit more.
Is this the case? I'm thinking we should change to Sub-Master but I need to 100% sure of the real life benefits.
The clustering you are referring to is solely for replicating the config across the proxy estate, this means that any filtering changes performed on the master server are picked up by all the site instances. The sub-master role is of use where you may have more then 2 servers, as in your case, and you would like one of the site instances to be able to take over the role of Master in the event of hardware failure.
We have in the passed had our site instances loose contact with the Master, however, every time this has happened (its rare and I've not had it happen in over a year) we have been able to fix by recreating the master/slave relationship thus saving a reboot.
When this happened the end users lost internet access until we rebooted the boxes, did everything carry on working for you or did you have to fix the relationship first.
Humm, no, everything continued working for us, the only impact was we were not pushing out policy changes from the Master. The Master/slave relation ship should not effect the running of the service though. As I understand how the Cluster works is, the salve servers periodically log on to the master via the management port over https (by default) and copy the config files off, and then install them to the running Webwasher instance. I guess if your Master config got horribly mangled it might be able to break the slaves, but that should not fix it self with a reboot.
Thanks again Tris most helpful,
I knew I had understood the documentation right, but just wanted to know if anyone else had seen this issue.
I think I might have to take the master down and see if this has the same impact as before, could be other issues going on.