On Jul 15, 2014, at 2:30 PM, Andrew <[email protected]> wrote:
> I guess it's possible that something along the way (NAT router or AWS maybe) > is actively involved in the socket and is hauling down what it considers to > be an inactive connection/translation. I recall another developer having issues with AWS closing idle sockets; but in that case I think it was clear to both sides of the connection what happened (the socket was explicitly closed, instead of packets just not arriving any longer.) So this seems different. > Is there any way of adjusting the Sync Gateway heartbeat without rebuilding > it to see if that's the case? The heartbeat interval is actually a parameter sent by the client in the request for the changes feed. It's exposed in the API as an optional property you can set in the CBLReplication's options dictionary, kCBLReplicatorOption_Heartbeat. The value is an NSNumber whose units are milliseconds. > Ah. But, more generally, isn't that exactly what CBL should be managing? By > definition, it's going to be running on devices that are likely to have less > than perfect connections. In that scenario, the passive WebSocket mode, where > CBL takes no responsibility for checking the connection or attempting its own > heartbeat, seems destined to go horribly wrong at some point! Or am I missing > something obvious? Yes, you're right. But I don't remember another case where there's trouble because of packets just getting dropped. Usually if there's a situation like that (like losing connectivity) the network interface state will change, which CBL will detect and automatically take the replication offline. I'm not sure what's going on here. --Jens -- You received this message because you are subscribed to the Google Groups "Couchbase Mobile" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/mobile-couchbase/81960133-7D6F-472D-B2C2-95010012E2B3%40couchbase.com. For more options, visit https://groups.google.com/d/optout.
