isReachable check hangs in a low network scenario

[tonymillion]

isReachable check hangs in a low network scenario

[pburleson]

I would move this check out of applicationDidFinishLaunchingWithOptions because it can take more than 10 seconds. You should schedule things to start happening in applicationDidFinishLaunchingWithOptions and get out of that method as fast as you can.

For instance, you could dispatch_async [reachHost startNotifier] and return from applicationDidFinishLaunchingWithOptions.

[AmirShaikh]

Thanks pburleson
I did try dispatch_async [reachHost startNotifier] but this didn't seem to make any difference to the reachable = [reachHost isReachable]; call which I can see it hanging on for a while before returning,

Should I be doing this somewhere else?

[tonymillion]

yes, unfortunately under poor network connectivity Reachability can hang in certain situations.

if you try something like:

reachable = NO;
reachHost = [Reachability reachabilityWithHostname:@"www.google.com"];

    reachHost.reachableBlock = ^(Reachability * reachability)
    {
        dispatch_async(dispatch_get_main_queue(), ^{
            reachable = YES;
        });
    };

    reachHost.unreachableBlock = ^(Reachability * reachability)
    {
        dispatch_async(dispatch_get_main_queue(), ^{
            reachable = NO;
        });
    };

dispatch_async(dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_DEFAULT, 0), ^{
   reachable = [reachHost isReachable];
   [reachHost startNotifier];
}

Then the initial detection will happen on a background thread, then the notifier will start up. Notice I put both the initial detection and the startNotifier in the async call.

[pburleson]

I'm trying to remember if Reachability has to be run on the main thread, but if it doesn't you should use the global queue instead of the main queue, as that continues to hang the main thread and will get you watchdog killed.

[tonymillion]

In addition, I'm curious why you need to run the initial isReachable?

Generally its better to try any network operations you were going to do, and catch them if they fail, queue them to be retried later, then when Reachability notifies you you have connectivity replay the queued operations.

[soleares]

isReachable shouldn't be called on app startup (on the main thread). Use the notification or blocks instead to update the reachability status. isReachable will block in low network situations which can result in the watchdog killing your app if it hangs for more than 20s. This will happen with Apple's reachability code too. I think it may talk about this in the readme or header in Apple's example.

[AmirShaikh]

I moved the isReachable check to a block and this works a lot better,

Tony, I agree it would be better not to call the initial isReachable but I'm making a data call to check for updates for my app as part of the launch, so i need to check for a connection.

Thanks for your replies all, will close if no further comments..