OK, here is Part 2 of the Q & A from MSCOM’s WebCast Week. These are from Tuesday’s session on Configuration and Management of Web Farms. The topics that Jeff Toews covered in this web cast included Configuration Management where he descried how we build out a server and our use of site-specific build scripts; Configuration Scripting that dove into how we remotely create and manage application pools, how we bind SSL certificates and registering assemblies in the Global Assembly Cache (GAC); Deploying Configurations to the Enterprise in which he described how we deploy URL Scan to a web server, how we get the content from a source location to a web cluster, how we synch the metabase within a cluster using iiscnfg.vbs; and he wrapped the session up with some Management Scripting information that described how we verify web server changes, and how we manage web service state.
Tuesday’s topic was: Configuration and Management of Web Farms
For an on demand viewing of this webcast go to:
Here is the recap of the questions:
A lot of what's been covered so far (including yesterday's webcast) focuses on the production deployment of mscom. Is that a mirrored staging deployment that things like private hotfixes are testing or content staged and tested? How close of a mirror is it if this exists or do you use a scaled down configuration for this?We use Pre-Production environments for content and hotfix testing. The server configuration is identical with the exception of the number of servers and some networking setting.
Do you actually set this (PeriodicRestartMemory) higher on x64 boxes? We do not use time based restarts. We use virtual memory settings. But the VM settings are identical.
How do you ensure that the app pools got created correctly in all the servers? Is there any script to verify the same ?This should work across multiple servers. ServerList.txt is a list of your servers. I used DefaulAppPool but you would want to use the name of your new AppPool. For /f %i in (Serverlist.txt) Do adsutil enum W3SVC/AppPools/defaultapppool -s:%i
Why do you specify the memory recycle? What will actually happen when the specified limit is reached?We want our AppPools to live as long as possible without impacting the applications within the AppPool. When the limit is reached, the new instance of that AppPool will be spun up and all new requests will be directed to the new AppPool while the old instance is being shutdown.
Can you use adsutil to start/stop iis-services for a web farm, like you have shown so far? For example to automize a shutdown script, for example do a maintanence (start the downsite, instead). If so, could you give a quick example?Here are all the options for adsutil. You can Stop and Start IIS. Utilizing the For Loop you can specify multiple servers. http://www.microsoft.com/windows2000/en/server/iis/default.asp?url=/windows2000/en/server/iis/htm/adminsamples/adsutil.htm.
Do you use application level test scripts to run against all the servers to ensure that the apppool setting is the same in all servers for the app ? is it there for our reference?We have monitoring tests which simulate the activities of an application that would alarm if the apppool settings are incorrect. We would normally use adsutil to get this information.
the URL for the webcast is (event it 1032283693)