site_archiver@lists.apple.com Delivered-To: macnetworkprog@lists.apple.com Thread-index: AQHUSrF8ywgf3mKwr0aQumTXXDCevqTuNAsAgAA0LAA= Thread-topic: A problem with big delays between 2 SCDynamicStore state events

On 13 Sep 2018, at 15:21, Quinn The Eskimo! <eskimo1@apple.com> wrote:

On 12 Sep 2018, at 16:58, Stéphane Sudre <ssudre@intego.com> wrote:

...

Has there been some changes (other than a bug/regression) that could explain the huge increase and that this is the expected behavior?

If, during this 90 second delay, you poll the network configuration, what do you see?

We’ve never really specified how long it takes for the network configuration to stabilise but there’s always been the possibility of a very long delay. For example, with IPv4 link-local addresses [1] we wait for DHCP to time out before we even start to set the process of getting a link-local address.

However, 90 seconds is clearly an annoyingly long time to wait.

Is this a regression relatively to 10.13.6?

After further investigations using the simple binary I was suggested to create for a bug report, this is actually not related to the SCDynamicStore events being delayed, it’s related to a libdispatch timer being fired with an excessive delay/leeway. As far as I can tell, it was due to using a global queue for the dispatch source. As there’s been some changes in GCD since OS X 10.8.5, this is a more reasonable explanation. And switching to using the main queue for the dispatch source solves the issue. _______________________________________________ Do not post admin requests to the list. They will be ignored. Macnetworkprog mailing list (Macnetworkprog@lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/macnetworkprog/site_archiver%40lists... This email sent to site_archiver@lists.apple.com