I am setting up transactional replication on my 2008R2 server to 2012 distribution server to 2012 Subscriber server. All the three servers are on FULL recovery mode. Log backup is taken every 15 minutes and full backup is taken daily everyday on all the three server databases used in replication. We want to use the subscriber as a reporting server. Publisher is our main OLTP database.
How can i get alert before my publisher's t-log space is full due to any issue in replication. I do not want my publisher to go down in any case. The t-logs are backed every 15 min on the publisher. Would that mean all the VLFs which are actively read by the log reader can be backed up when log backup is taken? what alerts should I use which will tell me that the log-space is getting full on publisher? Is there any script which can also stop replication/log reader agent when the log space is getting dangerously big on publisher?
I can remove replication and reestablish (with a new snapshot) when the replication is causing issue and building up logs on my publisher. But I dont want to wait for this situation to come (log space is full on publisher) and want to take action before only - like remove replication before only when log space reaches a certain percentage. How can i do that?
Also, what should be my strategy to stop and remove replication? I will be following Microsoft's documentation to remove replication but just not clear if I should stop the Log Reader agent and Distribution agent first before removing replication?