----- Original Message -----
On 02/12/2012 09:50 AM, Dan Kenigsberg wrote:
On Thu, Feb 09, 2012 at 03:23:19PM -0500, Ayal Baron wrote:
----- Original Message -----
-1
I agree that for messaging environment having a Message ID is a must because you sometimes don't have a particular target so when you get a response you need to know what this node is actually responding to.
The message ID could be composed with<FLOWID><MSGID> so you can reuse the field.
But that is all besides the point.
I understand that someone might find it fun to go on following the entire flow in the Engine and in VDSM. But I would like to hear an actual use case where someone would have actually benefited from this. As I see it having VSDM return the task ID with every response (and not just for async tasks) is a lot more useful and correct.
A generic debugging scenario as I see it.
- Something went wrong
- You go looking in the ENGINE log trying to figure out what
happend. 3. You see that ENGINE got SomeError. 4. Check to see if this error makes sense imagining that VDSM is always right and is a black box. 5. You did your digging and now you think that VDSM is as fault. 6. Go look for the call that failed. (If we returned the taskID it's pretty simple to find that call). 7. Look around the call to check VDSM state. 8. Profit.
There is never a point where you want to follow a whole flow call by call going back and forth, and even if you did having the VDSM taskID is a better anchor then flowID.
VDSM is built in a way that every call takes in to account the current state only. Debugging it with an engine flow mindset is just wrong and distracting. I see it doing more harm the good by reinforcing bad debugging practices.
I don't know about harm, but, today the engine logs every call and return value to and from vdsm. This means that all the info that is needed to follow a flow is already present in the engine log (which was not the case previously) so I believe that the flow id is redundant. In addition, instead of focusing on how to track a flow between components, we should focus on how to improve the engine log so that the users don't need to go to the hosts in the first place.
My biggest problem with it is that it changes each and every verb in the API and makes the log itself also more verbose and less readable.
The good thing about the currently suggested implementation
http://gerrit.ovirt.org/#patch,sidebyside,1221,6,vdsm/BindingXMLRPC.py
is that it (ab)uses an http header for carrying FlowID, thus keeping the formal API intact. FlowID is logged only on API entry point, so it would not clutter the logs too much.
if the FlowID isn't logged for logs which are part of the action, it will be easy to detect the entry point. but still wouldn't make it easy to grep all related logs to a flow
If the flow id is present in all lines it would make the log a lot less readable. Writing a script that accepts the flowid and 'grep's the 'flow' should be simple enough.
However, if this is to become (remotely) useful, the flowID should always contain an engine generated generation. The thing is, really complex debug scenarios which span multiple hosts never stem from the same flow, so the only added benefit here would be if we could at least know what happened before what from logs across multiple hosts. To achieve this we would need the engine to add a generation id to each unique flowID (the problem being that currently the flow ID would be determined by the UI and not by engine). Engine would have to generate such an ID for all internal flows as well!
I still have not heard even one good reason for having this though.