Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Stand Alone service limit on Windows Server 2012r2?
#1
We've been using Intraweb 14.1.13 for quite a while (I'm currently updating to 15), and we've run into some sort of hard limit on our AWS windows server on how many intraweb stand alone services we can run at the same time. Our customers are insurance agencies. We run a web application for them as a stand alone Intraweb service for each of our agencies. We run these on Amazon AWS windows server 2012 r2 ec2 instances. Each service connects to it's own database on an MS SQL 2014 community instance on the same server. Right now, memory is only at 57%, CPU never gets higher than 20%, plenty of empty diskspace. I can run 68 Intraweb services at once. When I add a new service for a new customer and try to start it, I get "Error 1053: The service did not respond to the start of control request in a timely fashion.". If I stop a currently running service, then I can start the new one. I created two Intraweb 'Hello World' services, one with version 14, and one with our new 15 version using Http.sys. I get the same result with those. With 68 services running, I can't start the Hello World service unless I stop a currently running service. However, I can start any other NON intraweb service. I have created a number of .Net services for various purposes on those servers and can start any number of them without problems.  

Does anyone else deploy their applications in this manner? (individual stand alone service per customer) Has anyone else run into this? The server has the resources to handle running quite a few more services. We have several AWS instances in different regions and I've run into this on three different servers.

I'd appreciate any thoughts anyone has on this crazy problem. Maybe I'm crazy, I'm not sure.

Thanks!
Reply
#2
Hello,

I am having the same problem with Http.sys.
I have the first server without a URL reservation and a few more, a few with a URL reservation.
There are two servers that abort with the same error message. I can restart it manually if I try multiple times.
All servers work according to the same scheme.
Some things are initialized between the start of main () and the call of TIWStartHSys :: Execute (), but the time between is less than 1 sec.

Unfortunately I have not been able to find a solution so far.

  Huh
Reply
#3
(08-05-2021, 04:18 PM)robAgencySoftware Wrote: We've been using Intraweb 14.1.13 for quite a while (I'm currently updating to 15), and we've run into some sort of hard limit on our AWS windows server on how many intraweb stand alone services we can run at the same time. Our customers are insurance agencies. We run a web application for them as a stand alone Intraweb service for each of our agencies. We run these on Amazon AWS windows server 2012 r2 ec2 instances. Each service connects to it's own database on an MS SQL 2014 community instance on the same server. Right now, memory is only at 57%, CPU never gets higher than 20%, plenty of empty diskspace. I can run 68 Intraweb services at once. When I add a new service for a new customer and try to start it, I get "Error 1053: The service did not respond to the start of control request in a timely fashion.". If I stop a currently running service, then I can start the new one. I created two Intraweb 'Hello World' services, one with version 14, and one with our new 15 version using Http.sys. I get the same result with those. With 68 services running, I can't start the Hello World service unless I stop a currently running service. However, I can start any other NON intraweb service. I have created a number of .Net services for various purposes on those servers and can start any number of them without problems.  

Does anyone else deploy their applications in this manner? (individual stand alone service per customer) Has anyone else run into this? The server has the resources to handle running quite a few more services. We have several AWS instances in different regions and I've run into this on three different servers.

I'd appreciate any thoughts anyone has on this crazy problem. Maybe I'm crazy, I'm not sure.

Thanks!

I'm assuming that all your other 68 instances are all Indy based services created with IW 14, is that correct?

I'll need to setup a test environment to try to replicate this, but it will take some time...
Reply
#4
(08-05-2021, 05:03 PM)JuergenS Wrote: Hello,

I am having the same problem with Http.sys.
I have the first server without a URL reservation and a few more, a few with a URL reservation.
There are two servers that abort with the same error message. I can restart it manually if I try multiple times.
All servers work according to the same scheme.
Some things are initialized between the start of main () and the call of TIWStartHSys :: Execute (), but the time between is less than 1 sec.

Unfortunately I have not been able to find a solution so far.

  Huh

I'm not sure if you are having the same problem. Service start can fail for different reasons. How many instances do you have?

Have you checked for the log files to see if there is anything in there?
Reply
#5
(08-05-2021, 10:27 PM)Alexandre Machado Wrote:
(08-05-2021, 05:03 PM)JuergenS Wrote: Hello,

I am having the same problem with Http.sys.
I have the first server without a URL reservation and a few more, a few with a URL reservation.
There are two servers that abort with the same error message. I can restart it manually if I try multiple times.
All servers work according to the same scheme.
Some things are initialized between the start of main () and the call of TIWStartHSys :: Execute (), but the time between is less than 1 sec.

Unfortunately I have not been able to find a solution so far.

  Huh

I'm not sure if you are having the same problem. Service start can fail for different reasons. How many instances do you have?

Have you checked for the log files to see if there is anything in there?

I have a total of 4 servers. I also checked all of the log files, including the Windows log files, but just this one message.
At first I suspected that sending email via the Indy components before calling TIWStartHSys :: Execute () was the cause.
But even if I only initialize the email components and send the email later, the problem occurs.
Reply
#6
(08-05-2021, 10:26 PM)Alexandre Machado Wrote:
(08-05-2021, 04:18 PM)robAgencySoftware Wrote: We've been using Intraweb 14.1.13 for quite a while (I'm currently updating to 15), and we've run into some sort of hard limit on our AWS windows server on how many intraweb stand alone services we can run at the same time. Our customers are insurance agencies. We run a web application for them as a stand alone Intraweb service for each of our agencies. We run these on Amazon AWS windows server 2012 r2 ec2 instances. Each service connects to it's own database on an MS SQL 2014 community instance on the same server. Right now, memory is only at 57%, CPU never gets higher than 20%, plenty of empty diskspace. I can run 68 Intraweb services at once. When I add a new service for a new customer and try to start it, I get "Error 1053: The service did not respond to the start of control request in a timely fashion.". If I stop a currently running service, then I can start the new one. I created two Intraweb 'Hello World' services, one with version 14, and one with our new 15 version using Http.sys. I get the same result with those. With 68 services running, I can't start the Hello World service unless I stop a currently running service. However, I can start any other NON intraweb service. I have created a number of .Net services for various purposes on those servers and can start any number of them without problems.  

Does anyone else deploy their applications in this manner? (individual stand alone service per customer) Has anyone else run into this? The server has the resources to handle running quite a few more services. We have several AWS instances in different regions and I've run into this on three different servers.

I'd appreciate any thoughts anyone has on this crazy problem. Maybe I'm crazy, I'm not sure.

Thanks!

I'm assuming that all your other 68 instances are all Indy based services created with IW 14, is that correct?

I'll need to setup a test environment to try to replicate this, but it will take some time...

That's correct. To be more specific, the dev environment consists of Rad Studio 10.2, Intraweb 14.1.14 (correction), CGDevTools 14.1.13, FastReports 5, and a few other things. I'm currently in the process of porting our application to Rad Studio 10.4.2, Intraweb 15.2.25, CGDevTools 15.2.23, FastReports, ect. 

Because we're working with the new version now, I did a Hello World test with IW15 (http.sys). Thrown into the mix, it behaves the same way unfortunately. I was hoping it would have something to do with the IW14 Indy server. The hello world doesn't connect to any DB, so that eliminates SQL Server as the problem. It (Hello World test) also only includes a single control, which is a TIWLabel. The problem doesn't appear to have anything to do with basic system resources like memory. I checked it early in the morning while the server was under fairly light load. Running on Windows Server 2012 r2. AWS EC2 instance. We've experienced it on 3 different servers now. Thanks very much for your time.
Reply
#7
(08-06-2021, 01:57 PM)robAgencySoftware Wrote:
(08-05-2021, 10:26 PM)Alexandre Machado Wrote:
(08-05-2021, 04:18 PM)robAgencySoftware Wrote: We've been using Intraweb 14.1.13 for quite a while (I'm currently updating to 15), and we've run into some sort of hard limit on our AWS windows server on how many intraweb stand alone services we can run at the same time. Our customers are insurance agencies. We run a web application for them as a stand alone Intraweb service for each of our agencies. We run these on Amazon AWS windows server 2012 r2 ec2 instances. Each service connects to it's own database on an MS SQL 2014 community instance on the same server. Right now, memory is only at 57%, CPU never gets higher than 20%, plenty of empty diskspace. I can run 68 Intraweb services at once. When I add a new service for a new customer and try to start it, I get "Error 1053: The service did not respond to the start of control request in a timely fashion.". If I stop a currently running service, then I can start the new one. I created two Intraweb 'Hello World' services, one with version 14, and one with our new 15 version using Http.sys. I get the same result with those. With 68 services running, I can't start the Hello World service unless I stop a currently running service. However, I can start any other NON intraweb service. I have created a number of .Net services for various purposes on those servers and can start any number of them without problems.  

Does anyone else deploy their applications in this manner? (individual stand alone service per customer) Has anyone else run into this? The server has the resources to handle running quite a few more services. We have several AWS instances in different regions and I've run into this on three different servers.

I'd appreciate any thoughts anyone has on this crazy problem. Maybe I'm crazy, I'm not sure.

Thanks!

I'm assuming that all your other 68 instances are all Indy based services created with IW 14, is that correct?

I'll need to setup a test environment to try to replicate this, but it will take some time...

That's correct. To be more specific, the dev environment consists of Rad Studio 10.2, Intraweb 14.1.14 (correction), CGDevTools 14.1.13, FastReports 5, and a few other things. I'm currently in the process of porting our application to Rad Studio 10.4.2, Intraweb 15.2.25, CGDevTools 15.2.23, FastReports, ect. 

Because we're working with the  new version now, I did a Hello World test with IW15 (http.sys). Thrown into the mix, it behaves the same way unfortunately. I was hoping it would have something to do with the IW14 Indy server. The hello world doesn't connect to any DB, so that eliminates SQL Server as the problem. It (Hello World test) also only includes a single control, which is a TIWLabel. The problem doesn't appear to have anything to do with basic system resources like memory. I checked it early in the morning while the server was under fairly light load. Running on Windows Server 2012 r2. AWS EC2 instance. We've experienced it on 3 different servers now. Thanks very much for your time.

I would also try just a simple windows service, not IW, maybe there is a windows limitation?
Reply
#8
Have you tried some other Delphi service, but not IntraWeb? It would be a good test. Myself I've already installed dozens of Delphi services on the same machine and never had any issues, but I don't remember going that far.
Reply
#9
I got 70 "Hello world" services running. No issues.

So, I guess there is something else involved in your case.

I'll try to add some more and see how it goes
Reply
#10
80 services running. All good. BTW, they are all Indy servers, built with the very latest IW 15.2.34.

Check it out

[Image: 80_IWServices.png]

So... in this case I'd suggest either:

1) You try to install the hello world app in a local server (I'm not aware of all limitations of AWS server but I'm sure they have some)

2) I can send you my test project and all bat files used to automate the process of install/uninstall/start/stop and you can try them on your same AWS server and see how it goes
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)