Tungsten Replicaiton Syncing Error

[Joseph Dique]

I have a tungsten replication version 3.0.1-64 setup installed in CentOS 6.5 Operating System. My setup is a star replication so all data is centralized in the main. My Replication has 14 nodes named from panel1 up to panel14, and 1 main named jjysrinnrepl. recently, I added a two new machine in the setup which is the panel15 and panel16. The lists below is the command used to add machine in the setup.

./tools/tpm configure activesystems \

--replication-user=tungsten \

--replication-password=M@st3rk3y1234 \

--install-directory=/opt/continuent \

--masters=panel1,panel2,panel3,panel4,panel5,panel6,panel7,panel8,panel9,panel10,panel11,panel12,panel13,panel14,panel15,panel16 \

--hub=jjysrinnrepl \

--hub-service=jjysrinnrepl \

--master-services=panel1,panel2,panel3,panel4,panel5,panel6,panel7,panel8,panel9,panel10,panel11,panel12,panel13,panel14,panel15,panel16 \

--topology=star \

--start

/tools/tpm update -fs

/apply_activeone_config_300.sh activeone_jjysrinntrading

In the 'masters=' and 'masters-services' portion, I mistakenly named one of the panels which is the panel4, I mistakenly use lower case the letter P whereas the original name of the machine is 'Panel4'. The two machine is successfully updated thou but the effect is that the panel4 cannot received data from the the other panels. I tried to use the trepctl services and trepctl -sevice panel4 status command to determine if it has a pending error but it turns it has no error and is up to date from panel4 to main and vice versa. The lists below is the dilemma in the panel4 now.

-All the data encoded from from the panel4 is succesfully replicated to the main and then to panels.

-All data encoded from the panels is successfully replicated to the main but is not replicated to the panel4.

-All data encoded from the main is successfully replicated to all the panels including the panel4

[Chris Parker]

Hi Joseph,

It sure sounds like a complex topology you have configured here!

It's going to be difficult for me to debug and understand what is going on given the v3 OSS version is no longer developed or maintained (You may have seen the announcements about this?)

But, given the 3 items you mentioned

-All the data encoded from from the panel4 is succesfully replicated to the main and then to panels.

This implies extraction from panel4 is ok and the appliers pulling from panel4 are ok

-All data encoded from the panels is successfully replicated to the main but is not replicated to the panel4.

This implies extraction on all the panels is working ok and that the apply to main is ok, this points therefore to the apply into panel4 being at fault

-All data encoded from the main is successfully replicated to all the panels including the panel4

However, the fact that data from main is coming into panel4 conflicts with the above statement.

Are there multiple applier services on each panel? What does "trepctl services" show? Have you looked at the logs to see if there are any clues there? If the panel4 service is only, have you simply tried restarting it? How did you fix the miss-configuration of the host name? Can you compare the configuration files for one of the other panels with panel4 to see if something has gone a miss?

Always good to remember that slaves PULL from the master, so if something isn't applying, the area to look at would be the applier, especially in your case where you know that the extraction portion is working.

Would you be interested in knowing a little more about the commercial release of the product? The commercial release is now at v6.1.3 and contains a considerable number of bug fixes and improvements since the OSS release, and on top of that, a subscription to the commercial release would provide you with full 24/7 support ensuring you get the best assistance to issues like this.  Our response rates for customers raising Urgent cases, over the last quarter averaged at 2 minutes.

I'd be happy to arrange a call to discuss this with you if you would be interested.

Thanks

Chris