[Bucardo-general] Bucardo design question
Michelle Sullivan
michelle at sorbs.net
Tue Aug 3 21:13:53 UTC 2010
Karl Denninger wrote:
>
> Slony can handle this by forcing a transaction separator - you can stick
> that into the CRON every 20 minutes or so, which keeps this problem from
> coming up.
>
> That being "missing" on Bucardo is a BIG problem IMHO, for exactly the
> reason you're running into.
Possibly - possibly not...
Looking at the delta table I see:
tablename | rowid | txntime | rowid2 |
rowid3 | rowid4
-----------+-----------+-------------------------------+----------+-------------------------------+--------
9830280 | 91726789 | 2010-07-30 01:44:48.570466+10 | |
|
9830315 | 91726789 | 2010-07-30 01:44:48.570466+10 | 21141601 |
|
9830375 | 21141601 | 2010-07-30 01:44:48.570466+10 | 91726789 |
2010-07-30 01:44:48.570466+10 |
9830363 | 90345828 | 2010-07-30 01:44:48.570466+10 | |
|
9830269 | 91726789 | 2010-07-30 01:44:48.570466+10 | 90345828 |
|
9830293 | 91726789 | 2010-07-30 01:44:48.570466+10 | 28 |
|
9830280 | 91726789 | 2010-07-30 01:44:48.570466+10 | |
|
9830233 | 123358616 | 2010-07-30 01:44:48.570466+10 | |
|
9830293 | 91726789 | 2010-07-30 01:44:48.570466+10 | 16436590 |
|
9830280 | 91726789 | 2010-07-30 01:44:48.570466+10 | |
|
9830233 | 123358617 | 2010-07-30 01:44:48.570466+10 | |
|
9830318 | 21143342 | 2010-07-30 03:21:37.6185+10 | |
|
14804208 | 1399248 | 2010-07-30 03:21:37.6185+10 | |
|
9830280 | 91733287 | 2010-07-30 03:21:37.6185+10 | |
|
9830315 | 91733287 | 2010-07-30 03:21:37.6185+10 | 21143342 |
|
9830375 | 21143342 | 2010-07-30 03:21:37.6185+10 | 91733287 |
2010-07-30 03:21:37.6185+10 |
9830363 | 90352323 | 2010-07-30 03:21:37.6185+10 | |
|
9830269 | 91733287 | 2010-07-30 03:21:37.6185+10 | 90352323 |
|
9830293 | 91733287 | 2010-07-30 03:21:37.6185+10 | 15224743 |
|
9830280 | 91733287 | 2010-07-30 03:21:37.6185+10 | |
|
9830233 | 123374693 | 2010-07-30 03:21:37.6185+10 | |
|
9830293 | 91733287 | 2010-07-30 03:21:37.6185+10 | 16945440 |
|
9830280 | 91733287 | 2010-07-30 03:21:37.6185+10 | |
|
9830233 | 123374694 | 2010-07-30 03:21:37.6185+10 | |
|
9830293 | 91733287 | 2010-07-30 03:21:37.6185+10 | 28 |
|
9830280 | 91733287 | 2010-07-30 03:21:37.6185+10 | |
|
9830233 | 123374695 | 2010-07-30 03:21:37.6185+10 | |
|
9830280 | 91735527 | 2010-07-30 03:53:18.504029+10 | |
|
9830315 | 91735527 | 2010-07-30 03:53:18.504029+10 | 19623147 |
|
9830375 | 19623147 | 2010-07-30 03:53:18.504029+10 | 91735527 |
2010-07-30 03:53:18.504029+10 |
9830363 | 90354562 | 2010-07-30 03:53:18.504029+10 | |
|
9830269 | 91735527 | 2010-07-30 03:53:18.504029+10 | 90354562 |
|
9830293 | 91735527 | 2010-07-30 03:53:18.504029+10 | 28 |
|
9830280 | 91735527 | 2010-07-30 03:53:18.504029+10 | |
|
9830233 | 123380254 | 2010-07-30 03:53:18.504029+10 | |
|
9830287 | 16947668 | 2010-07-30 03:53:18.504029+10 | |
|
14879660 | 3281688 | 2010-07-30 03:53:18.504029+10 | |
|
9830293 | 91735527 | 2010-07-30 03:53:18.504029+10 | 16947668 |
|
9830280 | 91735527 | 2010-07-30 03:53:18.504029+10 | |
|
9830233 | 123380257 | 2010-07-30 03:53:18.504029+10 | |
|
9830287 | 16947669 | 2010-07-30 03:53:18.504029+10 | |
|
14879660 | 3281689 | 2010-07-30 03:53:18.504029+10 | |
|
9830293 | 91735527 | 2010-07-30 03:53:18.504029+10 | 16947669 |
|
9830280 | 91735527 | 2010-07-30 03:53:18.504029+10 | |
|
9830233 | 123380258 | 2010-07-30 03:53:18.504029+10 | |
|
9830287 | 16947670 | 2010-07-30 03:53:18.504029+10 | |
|
14879660 | 3281690 | 2010-07-30 03:53:18.504029+10 | |
|
9830293 | 91735527 | 2010-07-30 03:53:18.504029+10 | 16947670 |
|
9830280 | 91735527 | 2010-07-30 03:53:18.504029+10 | |
|
etc..
This suggests that one could select a point in time and select all
transactions up to that... of course a modification to bucardo, but
perhaps select the transactions from the first to forward by 1 hour/1
day/whatever..
This is only a problem when trying to catchup, you can see by my data
I'm 4 days behind and with 28m rows I'm having an issue catching up...
Michelle
More information about the Bucardo-general
mailing list