rsyslogd crashed with SIGSEGV with juju-local configuration

[Martin Pitt]

installing juju-local pulls in rsyslog (at least on current xenial). When doing

- juju init
- edit ~/.juju/environments.yaml to set "local" as default
- juju boostrap

this creates a file /etc/rsyslog.d/25-juju-USERNAME-local.conf and restarts rsyslog. rsyslog then crashes like this with this configuration.

ProblemType: Crash
DistroRelease: Ubuntu 16.04
Package: rsyslog 8.14.0-2ubuntu2
ProcVersionSignature: Ubuntu 4.3.0-5.16-generic 4.3.3
Uname: Linux 4.3.0-5-generic x86_64
ApportVersion: 2.19.3-0ubuntu3
Architecture: amd64
CrashCounter: 1
Date: Thu Jan 14 12:18:18 2016
EcryptfsInUse: Yes
ExecutablePath: /usr/sbin/rsyslogd
ProcCmdline: /usr/sbin/rsyslogd -n
ProcEnviron:
 LANG=de_DE.UTF-8
 PATH=(custom, no user)
SegvAnalysis:
 Segfault happened at: 0x7efcf17d196e: mov 0x58(%rax),%rax
 PC (0x7efcf17d196e) ok
 source "0x58(%rax)" (0x00000058) not located in a known VMA region (needed readable region)!
 destination "%rax" ok
 Stack memory exhausted (SP below stack segment)
SegvReason: reading NULL VMA
Signal: 11
SourcePackage: rsyslog
StacktraceTop:
 ?? () from /usr/lib/rsyslog/lmnsd_gtls.so
 ?? () from /usr/lib/x86_64-linux-gnu/libgnutls-deb0.so.28
 ?? () from /usr/lib/x86_64-linux-gnu/libgnutls-deb0.so.28
 ?? () from /usr/lib/x86_64-linux-gnu/libgnutls-deb0.so.28
 ?? () from /usr/lib/x86_64-linux-gnu/libgnutls-deb0.so.28
Title: rsyslogd crashed with SIGSEGV
UpgradeStatus: No upgrade log present (probably fresh install)
UserGroups: adm

[Martin Pitt (pitti)]

Really shabby workaround for xenial: Change /lib/systemd/system/rsyslog.service like this:

[Service]
Type=oneshot
ExecStart=/bin/true
RemainAfterExit=yes

On that note, it would really be nice if juju-local would not require rsyslog. We have the systemd journal, so having rsyslog installed is just redundant (I normally don't have it).

[Apport retracing service (apport)]

StacktraceTop:
 gtlsClientCertCallback (session=<optimized out>, req_ca_rdn=<optimized out>, nreqs=<optimized out>, sign_algos=<optimized out>, sign_algos_length=<optimized out>, st=0x7efceed3e720) at nsd_gtls.c:254
 call_get_cert_callback () from /tmp/apport_sandbox_GOcefj/usr/lib/x86_64-linux-gnu/libgnutls-deb0.so.28
 _gnutls_server_select_cert () from /tmp/apport_sandbox_GOcefj/usr/lib/x86_64-linux-gnu/libgnutls-deb0.so.28
 remove_unwanted_ciphersuites () from /tmp/apport_sandbox_GOcefj/usr/lib/x86_64-linux-gnu/libgnutls-deb0.so.28
 _gnutls_server_select_suite () from /tmp/apport_sandbox_GOcefj/usr/lib/x86_64-linux-gnu/libgnutls-deb0.so.28

[Trent Lloyd (lathiat)]

Fixed upstream in 8.15:
https://github.com/rsyslog/rsyslog/issues/575

Latest release 8.16 is in Debian, perhaps we could just sync up to it?

[Launchpad Janitor (janitor)]

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

[Trent Lloyd (lathiat)]

Tested that patch, and it definitely solves the issue for me.

[Louis Bouchard (louis)]

This bug has been fixed by the latest upload of rsyslog : https://bugs.launchpad.net/ubuntu/+source/rsyslog/+bug/1539483. Marking it fix released