A stop gap solution - a better one may be to improve stats display on simulator-side.
Caps information is still accessible via the "show caps stats by user" and "show caps stats by cap" commands
For each service endpoint (e.g. posts to the xinventory service), a stat is available which shows the number of requests received and moving average per second
The full name is "service.<http-method>:<path>.requests (e.g. service.POST:/xinventory.requests)
- MemoryBuffer isn't seekable, so we can't log it. Log the string instead.
- Handle compressed streams
- Don't attempt to dump binary data. Either don't log it at all (if we know it's binary), or at least convert non-ASCII characters to ASCII.
- Log responses to HTTP requests
- Use the same log prefix for all of these log messages ("[LOGHTTP]"), to make them easy to see at a glance
- Increased the snippet length to 200 (80 doesn't show enough), and add "..." only if the message was actually truncated
Resolves http://opensimulator.org/mantis/view.php?id=6949
* Add an IsBlocked(string Key) method so it can be used more generically. (think.. if we want to rate limit login failures, we could have a call in the Login Service to IsBlocked(uuid.ToString()) and ignore the connection if it returns true, if IsBlocked returns false, we could run the login information and if the login fails we could run the Process method to count the login failures.
* Applied the XmlRpcBasicDOSProtector.cs to the login service as both an example, and good practice.
* Applied the BaseStreamHandlerBasicDOSProtector.cs to the friends service as an example of the DOS Protector on StreamHandlers
* Added CircularBuffer, used for CPU and Memory friendly rate monitoring.
* DosProtector has 2 states, 1. Just Check for blocked users and check general velocity, 2. Track velocity per user, It only jumps to 2 if it's getting a lot of requests, and state 1 is about as resource friendly as if it wasn't even there.
successfully tested, and I'm merging back those changes, which proved to
be good.
Revert "Revert "Cleared up much confusion in PollServiceRequestManager. Here's the history:""
This reverts commit fa2370b32e.
When Melanie added the web fetch inventory throttle to core, she made the long poll requests (EQs) effectively be handled on an active loop. All those requests, if they existed, were being constantly dequeued, checked for events (which most often they didn't have), and requeued again. This was an active loop thread on a 100ms cycle!
This fixes the issue. Now the inventory requests, if they aren't ready to be served, are placed directly back in the queue, but the long poll requests aren't placed there until there are events ready to be sent or timeout has been reached.
This puts the LongPollServiceWatcherThread back to 1sec cycle, as it was before.
This adds explicit cap poll handler supporting to the Caps classes rather than relying on callers to do the complicated coding.
Other refactoring was required to get logic into the right places to support this.
