Skip to content

[Discussion] ServiceController API needs some improvements #765

Description

@Fs00

I've been using the ServiceController class for some time now in an application of mine and I personally find that some features of the API could be improved. Here they are:

  • Timeout for stopping dependent services in Stop() is extremely long (30 secs).
    Which service takes even half of that time to stop itself? In my experience, when a stop request timed out instead of taking a couple of seconds to complete, it was because the dependent service somehow ignored it and refused to stop, causing my application to wait 30 seconds uselessly without having the ability to prevent it.
    I would suggest to reduce the timeout to something around 10 secs and/or add an overload to Stop that accepts a TimeSpan that specifies the timeout.
  • Stop method could be more convenient to use.
    The function DoStopSvc in this official WinAPI example checks if a stop is already pending before sending the stop control to the service, and in any case it waits for the service to stop before returning. Could ServiceController.Stop do the same? Waiting for a service to become stopped after a stop request is such a common use case that would be very convenient if Stop did it (it already does it when stopping dependent services).
  • System.ServiceProcess.TimeoutException message lacks the name of the service.
    The Stop method can raise a TimeoutException when a dependent service can't be stopped in time. The problem is that the exception doesn't tell which is the service that caused the timeout. Could this information be added in the message?

If you decide it's ok to proceed with these changes, I can take care of implementing them (for the last one I'd need some guidance on how to change exception messages resources).

PS: I wasn't sure whether to split this issue in multiple ones at this time since it is meant as a discussion on the topic.

Activity

  1. danmoseley commented on Feb 29, 2020

    @danmoseley
    Contributor

    We would certainly be interested in a PR that makes the exception(s) more useful.

  2. Fs00 commented on Mar 1, 2020

    @Fs00
    ContributorAuthor

    Ok, I'll look into it when I have some time. Can you point me in the right direction on how to change/add new exception message resources? (is some special operation needed?)
    And what about the other two proposals?

  3. danmoseley commented on Mar 1, 2020

    @danmoseley
    Contributor

    No special process for the exceptions..just a PR.

    @Anipik is owner of this area and could comment on the other two suggestions

  4. Anipik commented on Mar 2, 2020

    @Anipik
    Contributor

    @Fs00 I will take a look at the two proposals and update the issue.

  5. Fs00 commented on Apr 21, 2020

    @Fs00
    ContributorAuthor

    @Anipik Any news on those two proposals?

  6. Anipik commented on Apr 21, 2020

    @Anipik
    Contributor

    Any news on those two proposals?

    i will get back to u by tonight

  7. Anipik commented on Apr 22, 2020

    @Anipik
    Contributor

    and/or add an overload to Stop that accepts a TimeSpan that specifies the timeout.

    You can write api proposal for it and we will happy to take it. Here is goo example of an proposal #15725
    Changing the default value of 30 secs would be an issue as it might break some body else.

    Stop method could be more convenient to use.

    Feel free to throw up a pr to improve this.

  8. Fs00 commented on Apr 22, 2020

    @Fs00
    ContributorAuthor

    @Anipik After some thoughts, I've decided that it's better for consistency and compatibility to leave the Stop() method as-is. Instead, I've made a proposal for a new Stop overload with the improvements I'd like to see.

  9. added this to the Future milestone on Jun 24, 2020
  10. removed
    untriagedNew issue has not been triaged by the area owner
    on Jul 9, 2020
  11. added a commit that references this issue on Mar 25, 2021
  12. Fs00 commented on Jun 15, 2021

    @Fs00
    ContributorAuthor

    #33927 and #52519 solved the biggest pain points of this API, therefore I think that this issue can be closed now.
    Thank you all for your guidance! It's been great to be able to contribute 😃

  13. danmoseley commented on Jun 15, 2021

    @danmoseley
    Contributor

    @Fs00 thank you for your contributions! We have many other issues (mostly marked 'up-for-grabs') in case you are interested in continuing to contribute. Perhaps there is another bug you feel passionately about.

  14. ghost locked as resolved and limited conversation to collaborators on Jul 15, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions