[DRBD-user] Online verify causes kernel panic in XenServer 5.5

Lars Ellenberg lars.ellenberg at linbit.com
Tue Apr 6 17:17:40 CEST 2010

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


On Sun, Apr 04, 2010 at 01:44:24PM +0200, Alexander Thieme wrote:
> Hi List,
> 
> I am using XenServer 5.5 Update 2 with 2 servers in a XenPool and
> DRBD 8.3.5. Both nodes share a DRBD device and both are primary.
> 
> When I try to do an online verify of the resource one of the nodes crashes:
> Node 1, where I execute the command "drbdadm verify <resource>"
> continues to run, whereas node 2 crashes.

on freenode #drbd, you also mentioned http://pastebin.com/sixfY1JE
which I quote here:
   1.  block drbd1: role( Secondary -> Primary )
   2.  block drbd1: conn( StandAlone -> Unconnected )
   3.  block drbd1: Starting receiver thread (from drbd1_worker [4987])
   4.  block drbd1: receiver (re)started
   5.  block drbd1: conn( Unconnected -> WFConnection )
   6.  block drbd1: Handshake successful: Agreed network protocol version 91
   7.  block drbd1: conn( WFConnection -> WFReportParams )
   8.  BUG: warning at kernel/softirq.c:143/local_bh_enable() (Not tainted)
   9.  [<c010681a>] show_trace_log_lvl+0x1a/0x30
  10.  [<c0107052>] show_trace+0x12/0x20
  11.  [<c0107079>] dump_stack+0x19/0x20
  12.  [<c012cf08>] local_bh_enable+0xa8/0xb0
  13.  [<c02ba7ce>] lock_sock+0x8e/0xa0
  14.  [<c02ed8c8>] tcp_setsockopt+0xb8/0x3b0
  15.  [<c02b9fe2>] sock_common_setsockopt+0x22/0x30
  16.  [<f0718a7f>] drbd_worker+0x29f/0x480 [drbd]
  17.  [<f0734697>] drbd_thread_setup+0x137/0x1f0 [drbd]
  18.  [<c0103005>] kernel_thread_helper+0x5/0x10
  19.  =======================


Let me point you to 4.6 of the Xen Faq
http://wiki.xensource.com/xenwiki/XenFaq#head-a6ff59c593b136e2427534df391262f8b4ea7b1e

	4.6. I get "Badness in local_bh_enable at kernel/softirq.c" messages, why is this?

	This is fairly likely to be caused by a module compiled
	for native i386 rather than Xen. When building modules
	outside of the Xen build tree, use make ARCH=xen ....

	Alternative, this may be a driver that uses interrupt
	en/disabling instructions directly rather than the proper API....


Well, and in this case, I'd suggest this is simply a broken build.
And the below are results of the same.

"Works for me".

> I have a crash dump and I am not complete sure which lines are most
> interesting. Probably it is:
>    Call Trace:
>      [c01014a7] hypercall_page+0x4a7 (37: __HYPERVISOR_kexec_op)
>       c025f6c1  machine_kexec+0x21
>       c01d0069  sys_semctl+0x919
>       c01d0069  sys_semctl+0x919
>       c014a626  crash_kexec+0x66
>       c01dcf48  md5_update+0x88
>       c0106e8f  die+0x34f
>       c011a4eb  do_page_fault+0x68b
>       c0119e60  vmalloc_sync_all+0x4b0
>       c01060d3  error_code+0x2b
>       c01dcf48  md5_update+0x88
>       c01db7e5  update+0xa5
>       c0178da7  __kmalloc+0xc7
>       c02b9fe2  sock_common_setsockopt+0x22
>       c0130fd1  del_timer_sync+0x11
>       c0103005  kernel_thread_helper+0x5
> 
> When I change the hash function to sha1, I get the following:
> 
>    Call Trace:
>      [c01014a7] hypercall_page+0x4a7 (37: __HYPERVISOR_kexec_op)
>       c025f6c1  machine_kexec+0x21
>       c0320069  xfrm_state_find+0x7e9
>       c0320069  xfrm_state_find+0x7e9
>       c014a626  crash_kexec+0x66
>       c0327fa9  sha_transform+0x19
>       c0106e8f  die+0x34f
>       c011a4eb  do_page_fault+0x68b
>       c02ba72e  release_sock+0x9e
>       c02d3a07  __qdisc_run+0xd7
>       c0119e60  vmalloc_sync_all+0x4b0
>       c01060d3  error_code+0x2b
>       c0327fa9  sha_transform+0x19
>       c011c6a9  __kmap_atomic+0x189
>       c011c78f  kmap_atomic+0x1f
>       c01db7e5  update+0xa5
>       c0178da7  __kmalloc+0xc7
>       c02b9fe2  sock_common_setsockopt+0x22
>       c0130fd1  del_timer_sync+0x11
>       c0103005  kernel_thread_helper+0x5
> 
> 
> 
> If you need any further information, let me know. I am very
> interesting in getting this fixed.
> 
> Best regards,
> Alexander Thieme

-- 
: Lars Ellenberg
: LINBIT | Your Way to High Availability
: DRBD/HA support and consulting http://www.linbit.com

DRBD® and LINBIT® are registered trademarks of LINBIT, Austria.
__
please don't Cc me, but send to list   --   I'm subscribed



More information about the drbd-user mailing list