Sebastian Kaps napsal(a):
Hi Steve,
thanks for answering!
I have to correct myself:
it's Corosync v1.3.3 that comes with SLES11 SP1 HAE, not 1.1.5.
But I guess that's still too old.
On Thu, 10 May 2012 11:13:24 -0700, Steven Dake wrote:
This happens because synchronization doesn't complete. Synchronization
cannot complete for maybe 10-15 defects that are all fixed in the latest
stable release.
I will try to find a way to update both machines. In the meantime, is there
anything I can do when this happens? I can't kill the process in that state
and it seems to have a huge impact on the machine's performance.
BTW: I ran "corosync-cfgtool -s" with strace and it seems to have problems
writing to a file in /dev/shm:
Sebastian,
if it's really blocking in write, take a look to /dev/shm if there is
enough free space. If it's not, try to find out what is using that
space, try to delete files and see if it helps.
If there are only corosync files, maybe you've hit some error with tmp
buffers removal or maybe you've found a new bug.
----- snip ------
[...]
write(1, "Printing ring status.\n", 22) = 22
brk(0) = 0x603000
brk(0x624000) = 0x624000
socket(PF_FILE, SOCK_STREAM, 0) = 3
connect(3, {sa_family=AF_FILE, path=@"corosync.ipc"}, 110) = 0
open("/dev/shm/control_buffer-DwQiM7", O_RDWR|O_CREAT|O_EXCL, 0600) = 4
ftruncate(4, 8192) = 0
write(4,
"\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 4096
----- snip ------
That's the point where it hangs.
_______________________________________________
discuss mailing list
discuss@xxxxxxxxxxxx
http://lists.corosync.org/mailman/listinfo/discuss
