db2start error on 2-node cluster

[Robert Kane]

Good afternoon,

I'm trying to set up a clustered DB2 v7.1 database on Redhat Linux
7.3. I've followed the instructions (as far as I know) in the
accompanying documentation to set up the database on 2 nodes.

Running db2start on node 0 I get the following message:

11-19-2000 10:12:44 1 0 SQL6048N A communication error
occurred during START or STOP DATABASE MANAGER processing.
11-19-2000 10:12:45 0 0 SQL1063N DB2START processing was
successful.
SQL8011W One or more database partitions does not have a valid DB2
license key installed for the "DB2 Enterprise - Extended Edition"
product. Refer to db2diag.log for more information.
SQL6032W Start command processing was attempted on "2" node(s). "1"
node(s) were successfully started. "0" node(s) were already
started. "1" node(s) could not be started.

The database on node0 is started, but not the one on node1. The
license key issue is because the dual-Xeon processors are detected as
four, while the trial license is for 1 processor. The trial license is
"soft stop", which only gives a warning though and doesn't prevent the
database on node0 from running.

Running db2start on node 1 I got the message:

SQL1042C An unexpected system error occurred. SQLSTATE=58004.

Other miscellaneous obseverations:
1) Running "db2_all echo hi" produces expected output
2) There's a /database folder that was created on node0 but not node1
e) Running "db2admin start" on node1 produces a message that there is
no administration server. I tried to create one through the db2setup
utility, but it complains that ~db2as/sqllib directory is already
created (all the db2xxx users have nfs mounted home directories).

Is there anything obvious I've missed.

Thanks a bunch,
Robert Kane

[Fan Ruo Xin]

Rob,
Please check the db2 manual - Installation and Configuration Supplement. (You can reffer the Install
& Configure for UNIX part ).
If you followed each step, and still got the problem. Please post the db2diag.log messages when you
did DB2START.
Regards,
FRX

Robert Kane wrote:

Good afternoon,

I'm trying to set up a clustered DB2 v7.1 database on Redhat Linux
7.3. I've followed the instructions (as far as I know) in the
accompanying documentation to set up the database on 2 nodes.

Running db2start on node 0 I get the following message:

11-19-2000 10:12:44 1 0 SQL6048N A communication error
occurred during START or STOP DATABASE MANAGER processing.
11-19-2000 10:12:45 0 0 SQL1063N DB2START processing was
successful.
SQL8011W One or more database partitions does not have a valid DB2
license key installed for the "DB2 Enterprise - Extended Edition"
product. Refer to db2diag.log for more information.
SQL6032W Start command processing was attempted on "2" node(s). "1"
node(s) were successfully started. "0" node(s) were already
started. "1" node(s) could not be started.

The database on node0 is started, but not the one on node1. The
license key issue is because the dual-Xeon processors are detected as
four, while the trial license is for 1 processor. The trial license is
"soft stop", which only gives a warning though and doesn't prevent the
database on node0 from running.

Running db2start on node 1 I got the message:

SQL1042C An unexpected system error occurred. SQLSTATE=58004.

Other miscellaneous obseverations:
1) Running "db2_all echo hi" produces expected output
2) There's a /database folder that was created on node0 but not node1
e) Running "db2admin start" on node1 produces a message that there is
no administration server. I tried to create one through the db2setup
utility, but it complains that ~db2as/sqllib directory is already
created (all the db2xxx users have nfs mounted home directories).

Is there anything obvious I've missed.

Thanks a bunch,
Robert Kane

[Robert Kane]

Please check the db2 manual - Installation and Configuration Supplement. (You can reffer the Install
& Configure for UNIX part ).
If you followed each step, and still got the problem. Please post the db2diag.log messages when you
did DB2START.

Thanks for you response,

I've been following the DB2 EEE Installation and configuration guide for
Linux, and gone through the steps twice. The pre-install and install
stages work fine. The error occurs in the first step of the post-install
after creating the instance. In the pre-install phase, I have not
updated the _SHM_ID_BITS value in the kernel src, because it is present
in the kernel source I have (2.4.18-27.7.xsmp). From what I understand
though this is only recommend not required (and both kernels are the
same anyway) The software has been installed on both nodes.

I've created the instance as per the directions in the guide, and edited
the db2nodes.cfg file.
1) On one of the two nodes, I ran the db2setup program to create the
instance. The sqllib directory was created in /home/db2inst1 (shared via
NFS to both nodes)
2) I edited the db2nodes.cfg file so that it contains an entry for each
node.
3) As per the instructions I then attempted to run the db2start
command. The ~/sqllib/db2dump/db2diag.log file is appended the bottom.

Thanks again,
Rob

DB2DIAG.LOG
2000-11-20-15.41.59.080335 Instance:db2inst1 Node:000
PID:9667(db2star2) Appid:none
oper_system_services licRequestAccess Probe:1

SQL8007W There are "90" day(s) left in the evaluation period for the
product
"DB2 Enterprise - Extended Edition". For evaluation license terms and
conditions, refer to the IBM Evaluation Agreement in the EVALUATE.AGR
file,
located in the following directory: "/usr/IBMdb2/V7.1/Readme/C".
2000-11-20-15.41.59.088454 Instance:db2inst1 Node:000
PID:9582() Appid:none
base_sys_utilities sqleIssueStartStop Probe:90

7265 6376 2066 6169 6c75 7265 206f 6e20 recv failure on
6462 3273 7461 7274 206f 7220 6462 3273 db2start or db2s
746f 7020 7265 7375 6c74 2066 726f 6d20 top result from
6e6f 6465 2031 3a20 7263 203d 202d 3237 node 1: rc = -27
3130 330a 103.
2000-11-20-15.41.59.091405 Instance:db2inst1 Node:000
PID:9582() Appid:none
base_sys_utilities sqleIssueStartStop Probe:95

6462 3273 7461 7274 206f 7220 6462 3273 db2start or db2s
746f 7020 6661 696c 7572 6520 6672 6f6d top failure from
206e 6f64 6520 313a 2053 514c 3130 3432 node 1: SQL1042
4320 2041 6e20 756e 6578 7065 6374 6564 C An unexpected
2073 7973 7465 6d20 6572 726f 7220 6f63 system error oc
6375 7272 6564 2e0a curred..

[Fan Ruo Xin]

Rob,
When we met the error "A communication error ...", in most cases you need to check the file
INSTHOME/.rhosts.
But from the db2daig.log you posted here. It looks like this is not because of the .rhosts file.
I always
- when mount the NFS to the participated nodes, use the same mount point as that one on primary node.
- use the same username (userid) and groupname (groupid) for the three db2 users
default: db2inst1, db2as, db2fence1
- use the same service name and port number in /etc/services

If I were you, I would try to start the instance on two logical nodes (on the physical node 0) first. And
then work on adding the physical nodes.
If you check all these things, still get the problem. You can take a trace and then call IBM support.
Regards,
FRX
Robert Kane wrote:

Please check the db2 manual - Installation and Configuration Supplement. (You can reffer the Install
& Configure for UNIX part ).
If you followed each step, and still got the problem. Please post the db2diag.log messages when you
did DB2START.

Thanks for you response,

I've been following the DB2 EEE Installation and configuration guide for
Linux, and gone through the steps twice. The pre-install and install
stages work fine. The error occurs in the first step of the post-install
after creating the instance. In the pre-install phase, I have not
updated the _SHM_ID_BITS value in the kernel src, because it is present
in the kernel source I have (2.4.18-27.7.xsmp). From what I understand
though this is only recommend not required (and both kernels are the
same anyway) The software has been installed on both nodes.

I've created the instance as per the directions in the guide, and edited
the db2nodes.cfg file.
1) On one of the two nodes, I ran the db2setup program to create the
instance. The sqllib directory was created in /home/db2inst1 (shared via
NFS to both nodes)
2) I edited the db2nodes.cfg file so that it contains an entry for each
node.
3) As per the instructions I then attempted to run the db2start
command. The ~/sqllib/db2dump/db2diag.log file is appended the bottom.

Thanks again,
Rob

DB2DIAG.LOG

2000-11-20-15.41.59.080335 Instance:db2inst1 Node:000
PID:9667(db2star2) Appid:none
oper_system_services licRequestAccess Probe:1

SQL8007W There are "90" day(s) left in the evaluation period for the
product
"DB2 Enterprise - Extended Edition". For evaluation license terms and
conditions, refer to the IBM Evaluation Agreement in the EVALUATE.AGR
file,
located in the following directory: "/usr/IBMdb2/V7.1/Readme/C".

2000-11-20-15.41.59.088454 Instance:db2inst1 Node:000
PID:9582() Appid:none
base_sys_utilities sqleIssueStartStop Probe:90

7265 6376 2066 6169 6c75 7265 206f 6e20 recv failure on
6462 3273 7461 7274 206f 7220 6462 3273 db2start or db2s
746f 7020 7265 7375 6c74 2066 726f 6d20 top result from
6e6f 6465 2031 3a20 7263 203d 202d 3237 node 1: rc = -27
3130 330a 103.

2000-11-20-15.41.59.091405 Instance:db2inst1 Node:000
PID:9582() Appid:none
base_sys_utilities sqleIssueStartStop Probe:95

6462 3273 7461 7274 206f 7220 6462 3273 db2start or db2s
746f 7020 6661 696c 7572 6520 6672 6f6d top failure from
206e 6f64 6520 313a 2053 514c 3130 3432 node 1: SQL1042
4320 2041 6e20 756e 6578 7065 6374 6564 C An unexpected
2073 7973 7465 6d20 6572 726f 7220 6f63 system error oc
6375 7272 6564 2e0a curred..