[DRBD-user] DRBD Issues causing high server load

[Lars Ellenberg]

On Thu, May 03, 2018 at 08:18:06AM +0000, FAYE Jean Jacques Amos wrote:
> Hello,
> 
> We are experiencing problems with our DRBD replication system.
> 
> We upgraded from Ubuntu 12 to Ubuntu 14.04 (kernel 4.4.0-31-generic).
> 
> Our version of DRBD module is : 8.4.5
> The userland version is : 8.4.4

You posted to "drbd-dev", which is for coordination of development,
this question should go to drbd-user, so I moved it there.

As was already mentioned:
If you already do an upgrade,
why would you go for the oldest possible kernel in that upgrade?

Later kernels are available even for that old distribution you
chose to use now, I think currently they are at least at
4.4.0-121-generic...  first version to include the kernel fix for this
particular issue in the upstream stable 4.4 series was 4.4.33,
the first ubuntu 4.4 kernel including the 4.4.33 stable was likely
somewhere around 4.4.0-57, late 2016...

(Of course, it is your policy.
 I just Hope it prevents more hurt than it causes...)

> We often have disconnection problems between primaries and secondaries node.
> 
> Here is a log when the problem occurs :
> 
> Apr 28 05:30:28 XXXXXXXXXXXX kernel: [55510.085940] BUG: unable to handle kernel paging request at 0000000000001000

> Apr 28 05:30:28 XXXXXXXXXXXX kernel: [55510.093702]  [<ffffffff813e65b7>] ? copy_from_iter+0x2b7/0x2d0
> Apr 28 05:30:28 XXXXXXXXXXXX kernel: [55510.094415]  [<ffffffff8173f72d>] tcp_sendmsg+0x61d/0xad0

> Apr 28 05:30:28 XXXXXXXXXXXX kernel: [55510.097549]  [<ffffffffc0438c40>] drbd_send+0xc0/0x1b0 [drbd]
> Apr 28 05:30:28 XXXXXXXXXXXX kernel: [55510.098159]  [<ffffffffc043a641>] _drbd_no_send_page.isra.38+0x71/0xb0 [drbd]
> Apr 28 05:30:28 XXXXXXXXXXXX kernel: [55510.098704]  [<ffffffffc043ab6c>] drbd_send_dblock+0x33c/0x630 [drbd]
> Apr 28 05:30:28 XXXXXXXXXXXX kernel: [55510.099232]  [<ffffffff810bd784>] ? __wake_up+0x44/0x50
> Apr 28 05:30:28 XXXXXXXXXXXX kernel: [55510.099756]  [<ffffffffc0420bf4>] w_send_dblock+0x94/0x150 [drbd]

> Have you ever seen this kind of problem?

Yes.

> Do we have to downgrade the kernel version

No.
You should *upgrade*.

> or upgrade the drbd version ?

Yes, that as well.

> Thanks in advance for your help.

Cheers,

 :-)

-- 
: Lars Ellenberg
: LINBIT | Keeping the Digital World Running
: DRBD -- Heartbeat -- Corosync -- Pacemaker
: R&D, Integration, Ops, Consulting, Support

DRBD® and LINBIT® are registered trademarks of LINBIT