Activity overview
Latest activity by joe123
That is unfortunate. But bear in mind I am only referring to the command line version - if started from the UI, everything is just peachy, the correct (*only*) site is launched and profiled as expected - so the question is what form of black magic is RedGate.Profiler.UI.exe doing that Profile.exe isn't.
And assuming there really is no way to do that via command line, Is it possible to publish a list of messages the ui window will respond to so at least we can use the Win32 API / SendMessage to simulate someone pushing the relevant buttons?
Also.. whatever happened to the continuous profiling mode? http://www.red-gate.com/supportcenter/C ... figuration
no longer seems applicable, or at least I cannot find the " Continuous Profiling Configuration Tool." link / comments
That is unfortunate. But bear in mind I am only referring to the command line version - if started from the UI, everything is just peachy, the correct (*only*) site is launched and profiled as expe...
I think it is choosing the site it is choosing because it is the last in the list of sites with the same binding information..
Various sites have this binding info, verbatim:
<bindings>
<binding protocol="http" bindingInformation="*:80:" />
</bindings>
The currently active site i.e the one I want at the moment has the following information ( and in this case is the first in the list)
<bindings>
<binding protocol="http" bindingInformation="*:80:" />
<binding protocol="net.tcp" bindingInformation="808:*" />
<binding protocol="net.pipe" bindingInformation="*" />
<binding protocol="net.msmq" bindingInformation="localhost" />
<binding protocol="msmq.formatname" bindingInformation="localhost" />
</bindings>
Also all the deactivated sites have the attribute
serverAutoStart="false"
whereas the active one has
serverAutoStart="true"
Of course all this guessing could be avoided if the ants commandline start would just take a /site parameter that could pass in the site id ;P
Why is it that it is only supported through the ui / project files? Is the setup really that complex? / comments
I think it is choosing the site it is choosing because it is the last in the list of sites with the same binding information..
Various sites have this binding info, verbatim:
<bindings>
<binding pr...
Well to be fair they arent running simultaneously.
On a production server, yes I would imagine the host would be the prime differencing factor, but in localhost there basically ISN'T a nice way of dealing with this, so we stop one site and start the other as needed.
Is there a way to post semi private information? I could send you the <sites> section for the applicationHost.config so you could see what I meant, but I'm not too sure my employers would be amused with me posting said information in a public forum [image] / comments
Well to be fair they arent running simultaneously.
On a production server, yes I would imagine the host would be the prime differencing factor, but in localhost there basically ISN'T a nice way of ...
It returns *A* site name, but not the currently running site. ( in localhost in this particular case they all point to the same port so only one can be active at a time )
I would upload a screenshot but it would appear I am operating in semi retarded mode today and cant find the button to do so. / comments
It returns *A* site name, but not the currently running site. ( in localhost in this particular case they all point to the same port so only one can be active at a time )
I would upload a screensho...
Sorry I am not being clear. Ants would be installed on the server, so from its POV it would be a local machine.
However as the Continuous profiling mode still seems to be in beta, or at least not very well publicised, the issue and "solution" we are trying to implement is the following:
Problem:
We host and provide web services to various clients who have different peak times of use. We are constantly developing new bits and pieces, and we cannot allow the web service performance to slip, therefore regular stress testing is a necessary evil.
When we do said testing, we take one production server out of the load balancer, stress it with an in-house program, and then return it to the load balancer none the wiser.
Aim:
What we would like to be able to do, is from that in-house program send a message to the server to "start" ants, perform our load testing, collect the results, "stop" ants.
So basically the same process as before, but with profiling activated. In a way that is automated. I notice the memory profiler has an API for interacting with it, but the performance profiler doesn't seem to ( unless I am just looking in the wrong place).
My question is, given the tools that are available, how can I accomplish this goal? / comments
Sorry I am not being clear. Ants would be installed on the server, so from its POV it would be a local machine.
However as the Continuous profiling mode still seems to be in beta, or at least not v...
Ok, fair enough - but the issue is this is being launched from the command line, remotely. So how do I specify a url when the documentation specifically states that I need to use a project file. ( The save as from the main profiling application, I assume).
The project already has "localhost, restart on same port, port 80 " etc. saved, and as far as I can tell there is no way to override this information in the command line parameters (there isnt even a way to specify you want to profile IIS if you don't use a project file), so the question remains.. If i need to specify which site it refers to... HOW do I do that? / comments
Ok, fair enough - but the issue is this is being launched from the command line, remotely. So how do I specify a url when the documentation specifically states that I need to use a project file. ( ...
Continous / CommandLine profiling [v7.4.0.77]
Good news : So I finally managed to convince the wallet holders to splurge, and now we have paid licenses for ants - hooray!
Bad news: Now I have to put into practise the wodnerful things I said i...
Hi Dom,
Although your product team hasn't gotten around to it yet, I was able to try the patch Brian posted on another colleagues machine and it seems to be working OK again.
My red-gate evangelism campaign internally in the company is going well, and hopefully we should be able to place an order soon.
However we are also very interested in the continuous profiling aspect, which I notice is still in beta/EAP. Obviously on individual machines, the full blown experience is preferable, but am I correct in understanding the continuous profiling mode can be installed on a production server, and with history, if say a server experiences problems at 4am, the next day we would still be able to see the methods causing the issue the next day?
Does it offer the same level of detail (I.E. line-level timings? ) And what is the overall impact of actually having the profiler running permanently?
Finally, what are the licensing options? Is it paid on a per serv basis, or per the single developer assigned to work on N servers? / comments
Hi Dom,
Although your product team hasn't gotten around to it yet, I was able to try the patch Brian posted on another colleagues machine and it seems to be working OK again.
My red-gate evangelism...
There seems to have been a reversion in the release of 7.3, as they fixed this in the EAP if I remember - although the app pool launches in .net 4 mode, when restarting the IIS application on the same port, it does not seem to respect the original application pool's integration mode. As an aside, my trial expired while I was helping test the EAP, and now when the program starts (7.2) because the trial has expired it doesn't check for updates which means I cannot try the new version. I experienced this bug on a different machine.
Summary:
It appears APP 7.3 launches its IIS application pool with integrated security as opposed to classic, which prevent it from functioning, as dlls that require httpcontext.current return null. / comments
There seems to have been a reversion in the release of 7.3, as they fixed this in the EAP if I remember - although the app pool launches in .net 4 mode, when restarting the IIS application on the s...
Any further news on this? we have the same issue, and it makes Ants profiler next to useless. We can still attach to process but then you only have method-level timing. we have tried the early access preview builds and they too seem to suffer from the same issue, as does 6.3...
I ask because this was last looked at in amrch and there doesnt seem to have been any resolution... / comments
Any further news on this? we have the same issue, and it makes Ants profiler next to useless. We can still attach to process but then you only have method-level timing. we have tried the early acce...