[Bug 211033] sysutils/rsyslog8: CPU hangs at 100% with I/O error on stdout/stderr

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=211033

            Bug ID: 211033
           Summary: sysutils/rsyslog8: CPU hangs at 100%  with I/O error
                    on stdout/stderr
           Product: Ports & Packages
           Version: Latest
          Hardware: amd64
                OS: Any
            Status: New
          Severity: Affects Only Me
          Priority: ---
         Component: Individual Port(s)
          Assignee: brd at FreeBSD.org
          Reporter: xavi.garcia at gmail.com
             Flags: maintainer-feedback?(brd at FreeBSD.org)
          Assignee: brd at FreeBSD.org

rsyslogd hangs in an infinite loop trying to write a syslog record to a file
descriptor. 
truss -p pid outputs

write(1,"2016-07-06T10:01:23.061615+00:00"...,116) ERR#5 'Input/output error'
write(1,"2016-07-06T10:01:23.061615+00:00"...,116) ERR#5 'Input/output error'
write(1,"2016-07-06T10:01:23.061615+00:00"...,116) ERR#5 'Input/output error'
write(1,"2016-07-06T10:01:23.061615+00:00"...,116) ERR#5 'Input/output error'
write(1,"2016-07-06T10:01:23.061615+00:00"...,116) ERR#5 'Input/output error'
write(1,"2016-07-06T10:01:23.061615+00:00"...,116) ERR#5 'Input/output error'

It is not displayed here but the message written is exactly the same every time
(truss -s 1024 -p pid)


We assumed it was the same bug that was closed upstream on
https://github.com/rsyslog/rsyslog/issues/318 but we cannot reproduce it with
the proof of concept. Furthermore, building the binary with debug mode
(WITH_DEBUG=yes) stopped the bug from being triggered and we cannot get a
coredump to investigate.

We then forced the debug build with optimisations on (-O2) and we still cannot
reproduce the error.


We have seen this happening in fd 1 and fd 2.

The command `fstat -p pid`  returned

root     rsyslogd    1073    2 -         -         bad

that means bad file descriptor in stat(2) and errno.h (EBADF).

-- 
You are receiving this mail because:
You are the assignee for the bug.