|
Bugzilla – Full Text Bug Listing |
| Summary: | System hangs on shutdown after showing the shutdown of the service BIND | ||
|---|---|---|---|
| Product: | [openSUSE] SUSE LINUX 10.0 | Reporter: | Kurt Knopf <kurtknopf> |
| Component: | Other | Assignee: | Uwe Gansert <ug> |
| Status: | RESOLVED WORKSFORME | QA Contact: | E-mail List <qa-bugs> |
| Severity: | Critical | ||
| Priority: | P5 - None | ||
| Version: | Final | ||
| Target Milestone: | --- | ||
| Hardware: | i586 | ||
| OS: | SuSE Linux 10.0 | ||
| Whiteboard: | |||
| Found By: | Customer | Services Priority: | |
| Business Priority: | Blocker: | --- | |
| Marketing QA Status: | --- | IT Deployment: | --- |
| Attachments: | last 500 lines of /var/log/messages | ||
|
Description
Kurt Knopf
2005-10-30 12:59:47 UTC
When using larger logfiles, please use the attachment-feature of bugzilla. Please attach (!) 500 lines of your syslog: % tail -500 /var/log/messages > syslog.txt Thanks. Created attachment 56064 [details]
last 500 lines of /var/log/messages
Uwe: Does this help? I'm not sure what I could request here any further, the syslog does not contain much bind-messages either. Please ask about the specifics yourself. Thanks. please provide the output of "rcnamed status" on a running bind. two more infos please: 1. does "rnamed stop" work in a running system? 2. does it help to set RUN_PARALLEL=no in /etc/sysconfig/boot > (In reply to comment #5) > please provide the output of "rcnamed status" on a running bind. > The output is: igelserver:~ # rcnamed status Checking for nameserver BIND number of zones: 3 debug level: 0 xfers running: 0 xfers deferred: 0 soa queries in progress: 0 query logging is OFF recursive clients: 0/1000 tcp clients: 0/100 server is up and running running > (In reply to comment #6) > two more infos please: > > 1. does "rnamed stop" work in a running system? Yes without a problem. > 2. does it help to set RUN_PARALLEL=no in /etc/sysconfig/boot Changing RUN_PARALLEL to 'no' makes it even worse. The shutdown process still hangs after the message for the bind process but I am no longer able to initiate another shutdown task which would complete the shutdown. Maybe not the bind process is the problem but the task that would be shutdown after bind. with RUN_PARALLEL=yes, please shut down the system and if it hangs during the bind shutdown, check the process list if bind is still running. If it's still running, please do "strace -p" to the PID of the bind process. Maybe we can see where it hangs. > with RUN_PARALLEL=yes, please shut down the system and if it hangs during the
> bind shutdown, check the process list if bind is still running.
> If it's still running, please do "strace -p" to the PID of the bind process.
> Maybe we can see where it hangs.
Now it gets interesting. I tried to strace the running
named 5776 0.0 0.3 30324 3324 ? Ssl 22:10 0:00 /usr/sbin/named -t /var/lib/named -u named
but that returned nothing. But I found that the
'/bin/sh /etc/init.d/named stop'
task is still running as well. Stracing this process shows a 'waitpid(-1, '.
Then I tried to kill the '/bin/sh /etc/init.d/named stop' process which made the shutdown process continue successfully. So the question is why the stop command hangs on shutdown.
I hope you have any idea.
I'm digging in the dark, Can you please try to change the first line of the /etc/rc.d/named script from: "#! /bin/sh" to "#! /bin/sh -x" and then shutdown the system. The shell should print out what's going on and maybe we can see where it hangs. The shutdown hangs with the command
/usr/sbin/rndc status
Killing this process continues the shutdown process as well.
I tried to strace the /usr/sbin/rndc status (while being in the halted shutdown process) which lead to
---cut---
open("/usr/share/locale/POSIX/libisccc.cat", O_RDONLY) = -1 ENOENT (No such file or directory)
open("/usr/share/locale/POSIX/LC_MESSAGES/libisccc.cat", O_RDONLY) = -1 ENOENT (No such file or directory)
open("/usr/share/locale/POSIX/libisccc.cat", O_RDONLY) = -1 ENOENT (No such file or directory)
open("/usr/share/locale/POSIX/LC_MESSAGES/libisccc.cat", O_RDONLY) = -1 ENOENT (No such file or directory)
futex(0x400352f8, FUTEX_WAKE, 2147483647) = 0
open("/usr/share/locale/POSIX/libisc.cat", O_RDONLY) = -1 ENOENT (No such file or directory)
open("/usr/share/locale/POSIX/LC_MESSAGES/libisc.cat", O_RDONLY) = -1 ENOENT (No such file or directory)
open("/usr/share/locale/POSIX/libisc.cat", O_RDONLY) = -1 ENOENT (No such file or directory)
open("/usr/share/locale/POSIX/LC_MESSAGES/libisc.cat", O_RDONLY) = -1 ENOENT (No such file or directory)
futex(0x40184aa0, FUTEX_WAKE, 2147483647) = 0
futex(0x40184ac0, FUTEX_WAKE, 2147483647) = 0
futex(0x400352fc, FUTEX_WAKE, 2147483647) = 0
futex(0x8000b4f0, FUTEX_WAKE, 1) = 1
futex(0x8000b4ec, FUTEX_WAKE, 1) = 1
futex(0x8000b4b8, FUTEX_WAKE, 1) = 1
rt_sigaction(SIGHUP, {0x4016c330, ~[RTMIN RT_1], 0}, NULL, 8) = 0
rt_sigsuspend([]
--- cut ---
Seems like the rt_sigsuspend hangs the whole shutdown process.
does it change anything to set NAMED_RUN_CHROOTED="no" in /etc/sysconfig/named ? No. (In reply to comment #12) > does it change anything to set NAMED_RUN_CHROOTED="no" in /etc/sysconfig/named > ? No. Still hangs with the same problem. I tried a lot of stuff but I really can't reproduce that on any of my machines Is this really i586 or some other kind of hardware? I don't I can fix that as long as I can't reproduce it on my own. I can not reproduce it but I'll keep the problem in mind. Thanks for reporting. |