| Title: | Reliable Transaction Router |
| Moderator: | TALER::DESHMUKH |
| Created: | Tue Dec 12 1989 |
| Last Modified: | Thu Jun 05 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 695 |
| Total number of notes: | 2564 |
Hello, We have seen a network glitch where our sites take quite a while getting back in sync. The logs indicate a lot of traffic related to getting quorum back, etc. After all this activity, some transactions on the front end received a COMSTAUNO message, indicating something bad happened (sorry about the technical description, but the help says "Commit status unobtainable due to unrecoverable error"). This status wasn't returned until about ten minutes after the transaction was started. It seemed that the error was delivered about the same time that quorum was regained. My question is: Should we have to allow for a ten-minute delay in finding out the status of the transaction? Or is this a real RTR problem? Thanks for any advice, Roy Goodman
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 677.1 | When a COMSTAUNO is reported | FFRANC::DESHMUKH | Dipankar Deshmukh, RTR (US) Engg | Tue Feb 18 1997 00:57 | 12 |
Ten minutes is what RTR waits to find out what the final transaction status was, in order to report back to the client application. It is difficult to say whether this is an RTR bug, but there are bona fide circumstances under which a frontend node might not have a clue about how the transaction finally got resolved. In Citicorp's case, since the application is designed to cope with retransmitted transactions, perhaps the solution would be to repost them, and let the duplicate detection logic weed out the unnecessary ones. -dipu- | |||||