invalid replicate state change

I need some help activating two replicates on one of my instances. uname -a Linux psdb10rpr 2.6.32-431.17.1.el6.x86_64 #1 SMP Fri Apr 11 17:27:00 EDT 2014 x86_64 x86_64 x86_64 GNU/Linux crt01rf: onstat - IBM Informix Dynamic Server Version 12.10.FC4 -- On-Line -- Up 44 days 00:27:08 -- 4355376 Kbytes pscourts_rf: onstat - IBM Informix Dynamic Server Version 11.70.FC5 -- On-Line -- Up 100 days 19:22:35 -- 5364404 Kbytes I've created a replicateset with 4 replicates that replicate across 3 informix instances. All replicates on pscourts_rf and crt05sl started without incident. However 2 of the replicates on crt10rf are inactive. cdr start repl [replname] retunrs the 'invalid replicate state change' error. This is the script I used to define the replicates and I added them to the replset manually: cdr define repl --master=g_pscourtsrf --conflict=timestamp --scope=tran --ats --ris mentalgunrestrict_mental_case "mentalgunrestrict@g_pscourtsrf:informix.mental_case" "select * from mental_case" "mentalgunrestrict@g_crt10rf:informix.mental_case" "select * from mental_case" "mentalgunrestrict@g_crt05sl:informix.mental_case" "select * from mental_case" cdr define repl --master=g_pscourtsrf --conflict=timestamp --scope=tran --ats --ris mentalgunrestrict_mental_ident "mentalgunrestrict@g_pscourtsrf:informix.mental_ident" "select * from mental_ident" "mentalgunrestrict@g_crt10rf:informix.mental_ident" "select * from mental_ident" "mentalgunrestrict@g_crt05sl:informix.mental_ident" "select * from mental_ident" cdr define repl --master=g_pscourtsrf --conflict=timestamp --scope=tran --ats --ris mentalgunrestrict_ncic_posting "mentalgunrestrict@g_pscourtsrf:informix.ncic_posting" "select * from ncic_posting" "mentalgunrestrict@g_crt10rf:informix.ncic_posting" "select * from ncic_posting" "mentalgunrestrict@g_crt05sl:informix.ncic_posting" "select * from ncic_posting" cdr define repl --master=g_pscourtsrf --conflict=timestamp --scope=tran --ats --ris mentalgunrestrict_system_parameter "mentalgunrestrict@g_pscourtsrf:informix.system_parameter" "select * from system_parameter" "mentalgunrestrict@g_crt10rf:informix.system_parameter" "select * from system_parameter" "mentalgunrestrict@g_crt05sl:informix.system_parameter" "select * from system_parameter" crt10rf:cdr list replset mentalgunrestrict REPLICATE SET:mentalgunrestrict CURRENTLY DEFINED REPLICATES ------------------------------- REPLICATE: mentalgunrestrict_mental_case STATE: Inactive ON:g_crt10rf CONFLICT: Timestamp FREQUENCY: immediate QUEUE SIZE: 0 PARTICIPANT: mentalgunrestrict:informix.mental_case OPTIONS: transaction,ris,ats,fullrow REPLID: 6488126 / 0x63003e REPLMODE: PRIMARY ON:g_crt10rf APPLY-AS: INFORMIX ON:g_crt10rf REPLTYPE: Master REPLICATE: mentalgunrestrict_mental_ident STATE: Inactive ON:g_crt10rf CONFLICT: Timestamp FREQUENCY: immediate QUEUE SIZE: 0 PARTICIPANT: mentalgunrestrict:informix.mental_ident OPTIONS: transaction,ris,ats,fullrow REPLID: 6488127 / 0x63003f REPLMODE: PRIMARY ON:g_crt10rf APPLY-AS: INFORMIX ON:g_crt10rf REPLTYPE: Master REPLICATE: mentalgunrestrict_ncic_posting STATE: Active ON:g_crt10rf CONFLICT: Timestamp FREQUENCY: immediate QUEUE SIZE: 0 PARTICIPANT: mentalgunrestrict:informix.ncic_posting OPTIONS: transaction,ris,ats,fullrow REPLID: 6488128 / 0x630040 REPLMODE: PRIMARY ON:g_crt10rf APPLY-AS: INFORMIX ON:g_crt10rf REPLTYPE: Master REPLICATE: mentalgunrestrict_system_parameter STATE: Active ON:g_crt10rf CONFLICT: Timestamp FREQUENCY: immediate QUEUE SIZE: 0 PARTICIPANT: mentalgunrestrict:informix.system_parameter OPTIONS: transaction,ris,ats,fullrow REPLID: 6488129 / 0x630041 REPLMODE: PRIMARY ON:g_crt10rf APPLY-AS: INFORMIX ON:g_crt10rf REPLTYPE: Master

Sounds to me the primary key may be missing on those tables on the server that the rep is inactive. When you ran the command to create the replset if the pk was missing there should be a CDR error msg posted to the msg log at that timestamp. If not that then I suspect the some kind of mismatch between the schemas (you are using select * so the table schemas need to be the same). Regards, Mark

I suspect the some kind of mismatch between the schemas (you are using select * ) so the table schemas need to be the same including the primary key. When you created the repl set there would be a CDR error posted to the msg log on the server which the repl is Inactive. Regards, Mark