[Linux-cluster] some beginner questions regarding lock_gulmd

[Ashwin Sridharan]

Hi

I am a newbie trying to get lock_gulmd up and running 
on our SAN. I compiled the cluster code (from the
STABLE branch) using gcc 4.0.0 for a 2.6.13.4 kernel. 

For test purposes,  I started lock_gulmd 

as  ./lock_gulmd -s analysis-14 -n test and
got the following log   (analysis-14 is the name of
the machine I am running lock_gulmd on) :
-------------------------------------------------
Nov 13 18:36:56 analysis-14 lock_gulmd_LT[21685]: This
is cluster test 
Nov 13 18:36:56 analysis-14 lock_gulmd_core[21683]:
ERROR [src/core_io.c:1243] Failed to recv all of the
service login packet. -1:Unknown error 4294967295 
Nov 13 18:36:56 analysis-14 lock_gulmd_LT000[21685]:
ERROR [src/lock_io.c:576] Failed to receive login
reply. -104:104:Connection reset by peer 
Nov 13 18:36:57 analysis-14 lock_gulmd_main[21681]:
Forked lock_gulmd_LTPX. 
Nov 13 18:36:57 analysis-14 lock_gulmd_LTPX[21687]:
Starting lock_gulmd_LTPX 1.0.0. (built Jun 22 2005
17:05:20) Copyright (C) 2004 Red Hat, Inc.  All rights
reserved. 
Nov 13 18:36:57 analysis-14 lock_gulmd_LTPX[21687]: I
am running in Standard mode. 
Nov 13 18:36:57 analysis-14 lock_gulmd_LTPX[21687]: I
am (analysis-14) with ip (::ffff:10.64.228.114) 
Nov 13 18:36:57 analysis-14 lock_gulmd_LTPX[21687]:
This is cluster test 
Nov 13 18:36:57 analysis-14 lock_gulmd_core[21683]:
ERROR [src/core_io.c:1243] Failed to recv all of the
service login packet. -1:Unknown error 4294967295 
Nov 13 18:36:57 analysis-14 lock_gulmd_LTPX[21687]:
ERROR [src/ltpx_io.c:487] Failed to receive login
reply. -104:104:Connection reset by peer 

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

I believe there was a previous posting that this may
be due to incompatibility with gcc 4 but that the bugs
had been  fixed.  Odd thing is that if I run the
command 
as 

./lock_gulmd -s localhost -n test,

(i.e replace server name analysis-14 with localhost) 

I get no error messages, but a log saying that
lock_gulmd will not serve locks from this node and all
the gulm_tool commands time out. 

I've also tried the daemon from the code compiled with
gcc 3.2.2 and gcc 3.2.3  on the linux 2.6.13.4 kernel
with the same result. 

Oddly enough, when I did the same for a linux 2.6.12.4
kernel, the daemon worked perfectly (unfortunately due
to LUN issues I can't use that kernel).  Has anyone
experienced such issues or  have any insight into why
this may be happening (e.g. is it a kernel /gulm
incompatibility issue ) ?

Thanks
Ashwin


		
__________________________________ 
Yahoo! FareChase: Search multiple travel sites in one click.
http://farechase.yahoo.com