[Gluster-devel] Client side AFR during testing

[Snezhana Bekova]

  Hello,
I'm running glusterfs--mainline--2.5--patch-770 (glusterfs 1.3.9) and  
fuse-2.7.3glfs10 on 3 test machines with AFR with client side  
replication. My test setup is: 2 glusterfs servers and 3 glusterfs  
clients .The one machine is configured as client only and two are  
configured as server and client. While running tests with bonnie++ on  
the three clients (3 bonnies in parallel) I kill the one glusterfs  
server process and start again. After two or three test cycles one of  
the bonnies didn't finish at all with one of the errors:
"Can't open file ./Bonnie.17791.001"
"Can't open file 0000267hWNIpRcI"
"Expected 1024 files but only got 1039"

  The glusterfs client log messages are:
2008-05-27 15:44:28 E [afr.c:1239:afr_selfheal_setxattr_cbk] afr:  
(path=/Bonnie.1918/00002/0000267hWNIpRcI child=client-wks12) op_ret=-1  
op_errno=107
2008-05-27 15:44:28 W [client-protocol.c:332:client_protocol_xfer]  
client-wks1: not connected at the moment to submit frame type(1) op(31)
2008-05-27 15:44:28 E [client-protocol.c:2731:client_utimens_cbk]  
client-wks1: no proper reply from server, returning ENOTCONN
2008-05-27 15:44:28 E [afr.c:1279:afr_selfheal_utimens_cbk] afr:  
(path=/Bonnie.1918/00002/0000267hWNIpRcI child=client-wks1) op_ret=-1  
op_errno=107
2008-05-27 15:44:28 W [client-protocol.c:332:client_protocol_xfer]  
client-wks1: not connected at the moment to submit frame type(2) op(6)
2008-05-27 15:44:28 E [client-protocol.c:4270:client_unlock_cbk]  
client-wks1: no proper reply from server, returning ENOTCONN
2008-05-27 15:44:28 E [afr.c:1140:afr_selfheal_unlock_cbk] afr:  
(path=/Bonnie.1918/00002/0000267hWNIpRcI child=client-wks1) op_ret=-1  
op_errno=107
2008-05-27 15:44:28 E [afr.c:2177:afr_open] afr: self heal failed,  
returning EIO
2008-05-27 15:44:28 E [fuse-bridge.c:692:fuse_fd_cbk] glusterfs-fuse:  
113365: (12) /Bonnie.1918/00002/0000267hWNIpRcI => -1 (5)

  My config files on the two client and server mashines are:
cat /etc/glusterfs/glusterfs-server.vol
volume brick-local
  type storage/posix                  
  option directory /test       
end-volume

  volume p-locks
  type features/posix-locks
  subvolumes brick-local
  option mandatory on
end-volume

  volume server
  type protocol/server
  option transport-type tcp/server    
  subvolumes p-locks
  option auth.ip.p-locks.allow *
end-volume

cat /etc/glusterfs/glusterfs-client.vol
volume client-wks2
  type protocol/client
  option transport-type tcp/client
  option remote-host 10.0.0.2  
  option remote-subvolume p-locks
  option transport-timeout 5
end-volume

  volume client-wks1
  type protocol/client
  option transport-type tcp/client
  option remote-host 10.0.0.1  
  option remote-subvolume p-locks
  option transport-timeout 5
end-volume

  volume afr
  type cluster/afr
  subvolumes client-wks2 client-wks1
end-volume

  And this is config file on the third only glusterfs client mashine:
/etc/glusterfs/glusterfs-client.vol
volume client-wks1
  type protocol/client
  option transport-type tcp/client    
  option remote-host 10.0.0.1     
  option remote-subvolume p-locks 
  option transport-timeout 5
end-volume

  volume client-wks2
  type protocol/client
  option transport-type tcp/client    
  option remote-host 10.0.0.2       
  option remote-subvolume p-lock
  option transport-timeout 5
end-volume

  volume afr
 type cluster/afr
 subvolumes client-wks1 client-wks2
end-volume

Can you tell me what is wrong? Please advise and thanks in advance!

Thanks,
Snezhana

-------------------------------------

Хостинг от 2.60 лв/м | Домейни от 6.90 лв. | Сървъри, VPS от 42.00 лв/м с ДДС
12 GB място, Неограничен трафик, Безплатен домейн – 5.70 лв./м с ДДС!
17 GB място, 700 GB трафик, Безплатен домейн – 11.46 лв./м с ДДС!
 http://icn.bg/