[DRBD-user] High iowait when other node is down

[Mark Richardson]

  Hi, 

  I experience problems when only one node is up: 
Only a few writes to the filesystem put the iowait up to 99% and the volume (lvm) becomes unresponsive. 

  I've had the same problem when status is WFConnection and Standalone, this really defeats the purpose of replicating between two nodes if both of them need to be online, I think I am doing something wrong ? 

  my system: 
Linux stan 2.6.22-gentoo-r1 #1 SMP PREEMPT Fri Jul 13 21:06:48 CEST 2007 i686 Intel(R) Xeon(TM) CPU 3.06GHz GenuineIntel GNU/Linux 

  drbd version: Version: 8.0.5 (api:86) 

  LVM version:     2.02.09 (2006-08-17) 
Library version: 1.02.19 (2007-04-27) 
Driver version:  4.11.0 

  My config: 

  resource mirror0 { 
  protocol A; 
  # tried protocol C and B as well normally used C but thought 
  # perhaps this would help 
  startup { 
  } 
  disk { 
    on-io-error   detach; 
    # tried pass_on as well 
  } 
  net { 
  } 
  syncer { 
    rate 100M; 
    al-extents 257; 
  } 
  on stan { 
    device     /dev/drbd0; 
    disk       /dev/vg0/lv1; 
    address    192.168.1.101:7788; 
    meta-disk  internal; 
  } 
  on game01 { 
    device    /dev/drbd0; 
    disk      /dev/vg0/lv1; 
    address   192.168.1.102:7788; 
    meta-disk internal; 
  } 
} 

This problem first became apparent when trying to setup an volume in stand-alone mode as the other node could not be rebooted atm I figured I could configure one node and get the other one involved at an appropriate time, further testing showed the problem with the existing mirror as well (seems logical). 

  Sure hope I can resolve this issue as this only causes problems instead of preventing them. 

  Regards, 

  Mark 



----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linbit.com/pipermail/drbd-user/attachments/20070918/692786ed/attachment.htm>