- Change the maximum count of retry for web push notification (Default -> 5).
- In case of high load of subscribe server, the retries will be repeated many times.
- Because the retries occupy the default queue, maximum retry count should be reduced.
* Fix BackupService crashing when an attachment is missing
For various reasons such as admin error or out-of-sync media and
database backups, it might be possible for local attachments to be lost.
This commit allows the BackupService to continue its work even if some media
file is missing.
* Change error message
* Disable incorrect check for hidden services in Socket
Hidden services can only be accessed with an HTTP proxy, in which
case the host seen by the Socket class will be the proxy, not the
target host.
Hidden services are already filtered in `Request#initialize`.
* Use our Socket class to connect to HTTP proxies
Avoid the timeout logic being bypassed
* Add support for IP addresses in Request::Socket
* Refactor a bit, no need to keep the DNS resolver around
Use the smallest preset expires_in such that the new poll would
not expire before the old one.
In the typical case of a quick delete & redraft, this results in
using the same poll duration.
Fixes#10567
* Add request pool to improve delivery performance
Fix#7909
* Ensure connection is closed when exception interrupts execution
* Remove Timeout#timeout from socket connection
* Fix infinite retrial loop on HTTP::ConnectionError
* Close sockets on failure, reduce idle time to 90 seconds
* Add MAX_REQUEST_POOL_SIZE option to limit concurrent connections to the same server
* Use a shared pool size, 512 by default, to stay below open file limit
* Add some tests
* Add more tests
* Reduce MAX_IDLE_TIME from 90 to 30 seconds, reap every 30 seconds
* Use a shared pool that returns preferred connection but re-purposes other ones when needed
* Fix wrong connection being returned on subsequent calls within the same thread
* Reduce mutex calls on flushes from 2 to 1 and add test for reaping
Thibaut Girka