From be4193108513f02439a211a92fd80e0651f6721b Mon Sep 17 00:00:00 2001 From: Thomas Schwinge Date: Wed, 30 Nov 2011 21:21:45 +0100 Subject: IRC. --- open_issues/syslog.mdwn | 27 +++++++++++++++++++++++++++ 1 file changed, 27 insertions(+) (limited to 'open_issues/syslog.mdwn') diff --git a/open_issues/syslog.mdwn b/open_issues/syslog.mdwn index 5fec38b1..2e902698 100644 --- a/open_issues/syslog.mdwn +++ b/open_issues/syslog.mdwn @@ -43,3 +43,30 @@ IRC, freenode, #hurd, 2011-08-08 < youpi> shm should work with the latest libc < youpi> what won't is sysv sem < youpi> (i.e. semget) + + +IRC, OFTC, #debian-hurd, 2011-11-02: + + * pinotree sighs at #645790 :/ + pinotree: W.r.t. 645790 -- yeah, ``someone'' should finally + figure out what's going on with syslog. + http://lists.gnu.org/archive/html/bug-hurd/2008-07/msg00152.html + pinotree: And this... + http://lists.gnu.org/archive/html/bug-hurd/2007-02/msg00042.html + tschwinge: i did that 20 invocations tests recently, and + basically none of them has been logged + tschwinge: when i started playing with logger more, as result i + had some server that started taking all the cpu, followed by other + servers and in the end my ssh connection were dropped and i had nothing + to do (not even login from console) + pinotree: Sounds like ``fun''. Hopefully we can manage to + understand (and fix the underlying issue) why a simple syslog() + invocation can make the whole system instable. + tschwinge: to be honest, i got havoc in the system when i told + syslog to manually look for /dev/log (-u /dev/log), possibly alao when + telling to use a datagram socket (-d) + but even if a normal syslog() invocation does not cause havoc, + there's still the "lost messages" issue + Yep. What I've been doing ever since, is deinstall all + *syslog* packages. + This ``fixed'' all syslog() hangs. -- cgit v1.2.3