| Summary | Android - Fails to reconnect |
| Queue | gloox |
| Queue Version | SVN |
| Type | Bug |
| State | Unconfirmed |
| Priority | 3. High |
| Owners | |
| Requester | michael.ayles (at) melrosecomputing (dot) com |
| Created | 07/04/2017 (2349 days ago) |
| Due | |
| Updated | 07/06/2017 (2347 days ago) |
| Assigned | |
| Resolved |
tlsopensslbase.cpp
I've added:
void OpenSSLBase::cleanup()
{
if( !m_mutex.trylock() )
return;
if (m_ssl) {
SSL_shutdown(m_ssl);
SSL_clear(m_ssl);
}
m_secure = false;
m_valid = false;
m_mutex.unlock();
}
Also wasn't setting stream management after disconnect, re-connect
scenario which was causing another issue.
State ⇒ Unconfirmed
Priority ⇒ 3. High
Type ⇒ Bug
Summary ⇒ Android - Fails to reconnect
Queue ⇒ gloox
I'm experiencing issues in a reconnect scenario on Android. We're
using OpenSSL and connecting to ejabberd. If the device loses
connection (changes from WIFI -> Cell) and tries to reconnect it fails.
In the server logs it results in the following:
:handle_info:194 TLS error = SSL_do_handshake failed:
error:140760FC:SSL routines:SSL23_GET_CLIENT_HELLO:unknown protocol
Looking at the code for 1.0.20 it looks like we're still using the
same SSL object even though we have disconnected and tried to connect.
In order to try and patch the issue I made a change to the
disconnect logic to destroy the encryption object. This had the
desired effect in allowing us to connect once again with the server.
However I now get a different error.
GLOOX: AREA: 131072, <message from=test@test.com type='error'
id='2678af4260c9ef376bb893d0c632293424ad01fe00000013'><request
xmlns='urn:xmpp:receipts'/><error code='406'
type='modify'><not-acceptable
xmlns='urn:ietf:params:xml:ns:xmpp-stanzas'/></error><body>123</body></message>
So I'm assuming there is something persistent across sessions. Any
help would be appreciated.