2010/2/5 Gordan Bobic <gordan@xxxxxxxxxx>
nicolas prochazka wrote:How over-optimized is your build? Straight binary install or did you build it all with -O999 and suchlike?
Hello, For your information, glusterfs 3.0.2 still reboot my servers
Server 1 / Client 1 | Server2 / Client2 Client1 / Client2 in replicate mode to server 1 / 2 , no performance translator
Server 1 / Server 2 : 10.98.98.x /24 for gluster interface, 10.10.4.x/24 for external interface
Unplug ethernet cable from Server 1 ( port 1 10.98.98.x ) => after five minutes, servers reboot, no log, no dmesg.
Kernel 2.6.32.4 or 2.6.33r6 gentoo 64 bits distribution
No, i just make ./autogen.sh && ./configure && make && make install in git tree of glusterfs
fuse libs at the very least can be ruled out as a cause, since glfs 3.0.x doesn't use them, it connects straight to /dev/fuse, AFAIK.
xfs backend.
fuse 2.8.1
It's a little concerning that a process only communicating with kernel space through /dev/fuse can cause a reboot like that. One of the key points of fuse was to create enough separation to make that impossible...
Same server, same configuration, same tests, with gluster 2.9 => no reboot.
strange, isnt'it
Do you get a core dump? Is there a core.* file in the root directory ?
No core dump, i'm using a gentoo base distribution but system / reside in memory ( as a livecd ) with big memory system, so debugging is not the simple thing, i should install a gentoo 64bits this week.
Gordan
_______________________________________________
Gluster-devel mailing list
Gluster-devel@xxxxxxxxxx
http://lists.nongnu.org/mailman/listinfo/gluster-devel