I think that the logs could benefit from a couple of items:
1. I don't think that the log information being structured by User is very useful. I would rather see the logs, if they are tree-structured, be structured by Load Generator. It's difficult to debug with so many nodes on the log tree (potentially) and it's difficult to identify whether a certain trend is due to problems on a particular box.
2. I think it would be very beneficial if we could have a more traditional version of logs available to the user. This would just be a straight log file that allows for searching through the logs by time and text.
3. More choices to control the level of detail in the logs. I would like to be able to have more than just "none", "all", "all on one" and "errors on all": First, I'd like to see errors broken down by network errors, HTTP errors and other errors. If Users are the main partitioning point for logging information then the current scheme might make more sense; due to the fact that the amount of management for potentially tens of thousands of users is a bit too unwieldy. But if there was a logging scheme that was configured on a per load agent basis, then it would make more sense to have things fully configurable for the results fed from each load agent.
No replies to this topic
0 user(s) are reading this topic
0 members, 0 guests, 0 anonymous users