Re: ER setup cookbook

I just looked at this a bit more carefully and think that there might have been a problem. I should have checked it out a bit more carefully before I posted. Sorry. ER uses the 'group' name almost exclusivelly. Therefor on the define replicate command, the group name should be used instead of the INFORMIXSERVER. Madison Pruet wrote: > The following is the cookbook that I have refered to in previous > emails. We are putting in several example setup scripts in the 9.3 > documentation so hopefully this will be a much easier task than in the > past. The procedure for setting up the system in 7.31 and 9.20 is > similar to the one described. Hopefully this will help... > > ------------------------------------------------------------- > Summary: > > This documents intent is to assist customers in setting up ER > for the first time with 7.30 servers. It will detail how to > accomplish this by using the GUI on NT (Replication Manager) > or by using the Command Line Interface on UNIX or NT. > > -------------------------------------------------------------------------------- > > Detail: > Preparing for ER: > > - Databases on all instances involved should be set to logging > (unbuffereded logging is recommended) > - If using Conflict Resolution other than 'Ignore' for a > replicate - each table will need to have the CRCOLS columns > added. This can be done one of two ways: > - ALTER table tabname add CRCOLS; -- for existing tables > - CREATE table tabname (...) with CRCOLS; -- for new tables > - If planning to define the replicate using "select * from > tabname", are the tables identical on all instances defined > for this replicate? > - Any table defined in a replicate must have a PRIMARY KEY > constraint defined on at least one column. > > Setting up the SQLHOSTS file on UNIX: > > All machines where ER will be running on need to have a Trusted > relationship. To determine if your host is trusted by the other > host(s) in the ER system - use dbaccess -> connection -> connect, > and connect to the Group Name and the Servername of the other > host(s). > > Before using dbaccess to test this however - you need to set up > the SQLHOSTS file on all servers involved. Below is an example > of the SQLHOSTS file from the UNIX server 'unix1' before modified > for ER: > > ol_unix1shm onipcshm unix1 ol_unix1shm > ol_unix1tcp oltlitcp unix1 ol_unix1tcp > ol_unix2tcp oltlitcp unix2 ol_unix2tcp> > Also - the SQLHOSTS file on the other UNIX server 'unix2' would > look like this: > > ol_unix2shm onipcshm unix2 ol_unix2shm > ol_unix2tcp oltlitcp unix2 ol_unix2tcp > ol_unix1tcp oltlitcp unix1 ol_unix1tcp> > Currently, (based upon the files above) you can see that the two > servers already have connectivity to each other. This means that > the /etc/services file has entries for the TCP services listed in > the SQLHOSTS file, and there may also be entries in the /etc/hosts > or .rhost files for these servers. (Note: the TLI connection > method is only supported on certain platforms - please check the > Machine Specific Notes for you engine to find out if your platform > support TLI, or SOC connections.) > > The following modifications will need to be made to enable ER to > be defined on each of these servers. > > The Unix1 server: > unix1grp group - - i=1 > ol_unix1tcp oltlitcp unix1 ol_unix1tcp g=unix1grp > ol_unix1shm onipcshm unix1 ol_unix1shm g=unix1grp > > The Unix2 server: > unix2grp group - - i=2 > ol_unix2tcp oltlitcp unix2 ol_unix2tcp g=unix2grp > ol_unix2shm onipcshm unix2 ol_unix2shm g=unix2grp > > Verify that the INFORMIXSERVER environment variable is set for the > network > connection and not the shared memory connection. For this example set > INFORMIXSERVER to "ol_unix1tcp". Also, pre-9.20, set the DBSERVERNAME > in the > onconfig file for both servers to the TCP connection. Please check out > bug > #113269 for more info on this. > > Setting up the SQLHOSTS registry entries on NT: > > There is a demo program in $INFORMIXDIR/demo named ershed.exe - > this is the SQLHOSTS editor. This tool can read an SQLHOSTS > (that has been copied from a UNIX system) and write it into the > registry on NT. If you are just setting up NT boxes however, > this tool can still be helpful - as you and create a standard > SQLHOSTS file (as you would on UNIX) and use it write all the > registry entries needed. If you feel uncomfortable using this > tool - then the release notes on NT and UNIX detail two other > ways to accomplish the same task - 1) by using the IECC client, > and 2) by adding the entries directly to the registry. > > After the registry has a group entry for each server - use the > SetNet32 utility to populate the host information with a username > and password for each host. Then use I-Login to make a connection > to each host - to ensure connectivity. If you have an engine > installed on NT - you can use dbaccess to insure this connectivity. > > Using the Command Line on UNIX or NT: > > 1 - Define the server: > > UNIX: cdr define server -c unix1grp -s rootdbs -r rootdbs \\ > -A /tmp -R /tmp -I unix1grp > > NT: cdr define server -c ai_airprt1 -A /tmp -R /tmp \\ > -I ai_airprt1 -S ai_unix1grp > > 2 - Define the replicate: > > Update Anywhere Syntax: > cdr define repl -c unix1grp -C"timestamp" -S tran -A repA \\ > "database@ol_unix1tcp:informix.tablename" "select * from tablename" \\ > "database@ol_unix2tcp:informix.tablename" "select * from tablename" > > Primary/Target Syntax: > cdr define repl -c unix1grp -C"timestamp" -S tran -A repA \\ > "P database@ol_unix1tcp:informix.tablename" "select * from tablename" \\ > "R database@ol_unix2tcp:informix.tablename" "select * from tablename" > > 3 - Start the replicate > cdr start repl repA unix1grp unix2grp > > Using the Replication Manager on NT: > > Refer to Chapters 7 and 8 of the Guide to Informix Enterprise > Replication version 7.3. These chapters go into great detail on > defining the server and starting replication using the GUI tool.