hi,

i'm testing the aoe/vblade as a backend for mdadm/raid10 with the  
lvm/ocfs2 on the top.
during i/o tests i'm observing ranfom failures of /dev/etherd/e*.* devices.

in the worst case the 'dd if=/dev/zero of=/dev/etherd/e... bs=1M  
count=4096' leads
to machine reboot without any logs but i have a reduced non-rebootable  
testcase
that could be related.

here're steps to reproduce:

on the first console:

# rmmod aoe
# modprobe aoe aoe_iflist=lo
# ls -lh /home/testfile
-rw-r--r-- 1 root root 8,0G 07-26 16:11 /home/testfile
# vblade 666 0 lo /home/testfile
pid 4589: e666.0, 16777216 sectors O_RDWR

on the second console:

# aoe-discover && aoe-stat
e666.0         8.589GB     lo up
# dd if=/dev/zero of=/dev/etherd/e666.0 bs=1M count=1024
1024+0 records in
1024+0 records out
1073741824 bytes (1.1 GB) copied, 6.66507 s, 161 MB/s

and one more copy...

# dd if=/dev/zero of=/dev/etherd/e666.0 bs=1M count=1024
<freeze...>

now back to first console and there's a message from vblade:

read network: Bad address

the end of vblade strace log contains:

(...)
read(4, 0x603fdc, 65536)                = -1 EFAULT (Bad address)
dup(2)                                  = 5
fcntl(5, F_GETFL)                       = 0x8002 (flags O_RDWR|O_LARGEFILE)
fstat(5, {st_mode=S_IFCHR|0620, st_rdev=makedev(136, 1), ...}) = 0
mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) =  
0x7f6b12f1d000
lseek(5, 0, SEEK_CUR)                   = -1 ESPIPE (Illegal seek)
write(5, "read network: Bad address\n", 26read network: Bad address) = 26


adding the '-s' switch to vblade seems to workaround this error but  
impacts performance.
could you please look at this error?

i'm using following software:

aoetools-30-2.x86_64
vblade-20-1.x86_64
kernel-2.6.34.1-3.x86_64
glibc-2.12-4.x86_64
gcc-4.5.0-9.x86_64

BR,
Pawel.

------------------------------------------------------------------------------
The Palm PDK Hot Apps Program offers developers who use the
Plug-In Development Kit to bring their C/C++ apps to Palm for a share 
of $1 Million in cash or HP Products. Visit us here for more details:
http://ad.doubleclick.net/clk;226879339;13503038;l?
http://clk.atdmt.com/CRS/go/247765532/direct/01/
_______________________________________________
Aoetools-discuss mailing list
Aoetools-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/aoetools-discuss

Reply via email to