Random segfaults on __pthread_create_2_1

[jclariana]

We are experiencing random segfaults on our pdnsd daemon.

System versions:

Ubuntu 16.04.1 LTS
Arch: 64bits
pdnsd version: 1.2.9a-par-3
libc6 version: 2.23-0ubuntu3
kernel version: Linux api 4.4.0-38-generic #57-Ubuntu SMP Tue Sep 6 15:42:33 UTC 2016 x86_64 x86_64 x86_64 GNU/Linux

In syslog:

Sep 24 07:05:35 api kernel: [84418.840445] pdnsd[32156]: segfault at 7f18f395cd13 ip 00007f18f456decf sp 00007f18f2158bc0 error 4 in libpthread-2.23.so[7f18f4566000+18000]

I enabled core dump, apport report is attached to this issue.
When trying to backtrace the segfault:

Core was generated by `/usr/sbin/pdnsd --daemon -p /var/run/pdnsd.pid'.
Program terminated with signal SIGSEGV, Segmentation fault.
#0 __pthread_create_2_1 (newthread=<optimized out>, attr=<optimized out>, start_routine=<optimized out>, arg=<optimized out>) at pthread_create.c:713
713 pthread_create.c: No such file or directory.
[Current thread is 1 (Thread 0x7f34be48a700 (LWP 2991))]
(gdb) bt
#0 __pthread_create_2_1 (newthread=<optimized out>, attr=<optimized out>, start_routine=<optimized out>, arg=<optimized out>) at pthread_create.c:713
#1 0x000000000040ece5 in ?? ()
#2 0x00007f34c089e6fa in start_thread (arg=0x7f34be48a700) at pthread_create.c:333
#3 0x00007f34c05d4b5d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:109

pdnsd config file:

// Read the pdnsd.conf(5) manpage for an explanation of the options.
/* Note: this file is overriden by automatic config files when
   /etc/default/pdnsd AUTO_MODE is set and that
   /usr/share/pdnsd/pdnsd-$AUTO_MODE.conf exists
 */
global {
    perm_cache = 2048;
    cache_dir = "/var/cache/pdnsd";
    run_as = "pdnsd";
    server_ip = 127.0.0.1;
    status_ctl = on;
    paranoid = on;
    min_ttl = 15m; // Retain cached entries at least 15m.
    max_ttl = 1w;
    timeout = 10; // Global timeout option (10 seconds).
    neg_rrs_pol = on;
    par_queries = 1;
}

server {
    label="resolvconf";
    file="/var/run/resolvconf/interface/eth0.dhclient";
    proxy_only=on;
}

source {
    owner=localhost;
    file="/etc/hosts";
}

rr {
    name=localhost;
    reverse=on;
    a=127.0.0.1;
    owner=localhost;
    soa=localhost,root.localhost,42,86400,900,86400,86400;
}

/* vim:set ft=c: */

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Krzysztof Warunek (kwarunek)]

I experience similar problem. Crash happens after some time ~20h.

My configs:

/etc/pdnsd.conf:
# -- begin pdnsd.conf --
global {
    perm_cache = 32768;
    cache_dir = "/var/cache/pdnsd";
    run_as = "pdnsd";
    server_ip = 127.0.0.1; // Use eth0 here if you want to allow other
                            // machines on your network to query pdnsd.
    status_ctl = on;
    paranoid = on;
    min_ttl = 0s; // We want to honour DNS TTL
    max_ttl = 1w; // One week.
    neg_ttl = 5s;
    timeout = 10; // Global timeout option (10 seconds).
}

server {
    label=OnetDNS;
    ip=10.10.44.11;
    ip=10.10.44.21;
    timeout=2;
    purge_cache=off;
    proxy_only=on;
    uptest=ping;
    ping_timeout=10;
    randomize_servers=on;
}

rr {
    name = localhost;
    reverse = on;
    a = 127.0.0.1;
    owner = localhost;
    soa = localhost,root.localhost,42,86400,900,86400,86400;
}

source {
    owner = localhost;
    file = "/etc/hosts";
    serve_aliases=on;
}

source {
    owner = localhost;
    file = "/etc/hosts.mydomain";
    serve_aliases=on;
}
include {
    file = /etc/pdnsd.d/locdom.conf;
}
# -- end pdnsd.conf --

/etc/pdnsd.d/locdom.conf
# -- begin locdom.conf --
rr {
    name = onetapi.pl;
    ns = localhost;
    soa = localhost,root.localhost,42,86400,900,86400,86400;
    a = 10.177.0.1;
}

rr {
        name = *.locdom.com;
        cname = locdom.com;
}
# -- end locdom.conf --

[Krzysztof Warunek (kwarunek)]

update, crash happens randomly, the assumption of 20h is false.