On Tue, Apr 21, 2015 at 12:58 PM, Howard Chu <hyc(a)symas.com> wrote:
> Full_Name: Andy
> Version: 2.4.30
> OS: Debian Linux (Jessie)
> URL: ftp://ftp.openldap.org/incoming/
> Submission from: (NULL) (188.8.131.52)
> When connecting over TLS to an LDAP server with high packet loss, my calls to
> ldap_sasl_bind_s() block forever some of the time despite have set a 10s limit
> on all of LDAP_OPT_NETWORK_TIMEOUT, LDAP_OPT_TIMEOUT and LDAP_OPT_TIMELIMIT.
> I see the following stack trace on the deadlocked process.
> @ 0x7f2ff0c0f8d0 (unknown)
> @ 0x7f2ff0c0eac0 __libc_read
> @ 0x5cdc68 sb_debug_read
> @ 0x5c26f5 tlso_bio_read
> @ 0x451b64 BIO_read
> @ 0x49bf86 ssl3_read_n
> @ 0x49d52b ssl3_get_record
> @ 0x49caea ssl3_read_bytes
> @ 0x496f61 ssl3_get_message
> @ 0x496944 ssl3_get_finished
> @ 0x5e4aab ssl3_connect
> @ 0x5e164b ssl23_connect
> @ 0x5c1b3c tlso_session_connect
> @ 0x5b6773 ldap_int_tls_start
> @ 0x5b3a77 ldap_int_open_connection
> @ 0x5a730e ldap_new_connection
> @ 0x5b3365 ldap_open_defconn
> @ 0x5a68e3 ldap_send_initial_request
> @ 0x5a4c1e ldap_sasl_bind
> @ 0x5a4d99 ldap_sasl_bind_s
> Inspecting the source, I see that SSL_connect() is called from
> tlso_session_connect(). Does openldap attempt to pass timeout information to
> Sorry for the vague bug report. The issue is non-deterministic so it is hard to
> provide too much information. Please let me know what I can do to be more
> I was able to find an old posting (from 2006) reporting a similar issue. I'm
> uncertain if the root cause is the same or not though.
There is no API for setting a timeout in the TLS library.
Thanks for the pointer. Feel free to close this issue.