[DRBD-user] drbd, blocked processes in D state, send-Q, recv-Q

Lars Ellenberg lars.ellenberg at linbit.com
Fri Oct 10 22:05:12 CEST 2014

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


On Mon, Aug 25, 2014 at 02:36:33PM +0200, Roland Friedwagner wrote:
> Hello,
> 
> blocking is a normal behavor when you replication link TCP queues 
> becomes full.
> 
> You may increase the linux kernel defaults like this in /etc/sysctl.conf
>   net.core.rmem_default = 16777216
>   net.core.wmem_default = 16777216
>   net.core.rmem_max = 33554432
>   net.core.wmem_max = 33554432
>   net.ipv4.tcp_rmem = 131072 16777216 33554432
>   net.ipv4.tcp_wmem = 131072 16777216 33554432


You do realize that the setting above is in units of Byte per socket,
but the setting below:

>   net.ipv4.tcp_mem  = 131072 16777216 33554432

This is in global *pages*, used by anything in the TCP stack.

You really have 128 GiB of RAM to spend on your TCP stack alone??
That may well be (I've worked on systems with > 1 TiB RAM...)
But it does not seem likely...

Read: do not simply copy'n'paste from ubuntu-users forum or similar...
Check what those settings do, and if the proposed setting is plausible!

-- 
: 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