tag:blogger.com,1999:blog-826910957631479390.post6094063520370836705..comments2024-03-28T03:15:27.747-07:00Comments on Evol Monkey: PostgreSQL Bi-Directional ReplicationEvol Monkeyhttp://www.blogger.com/profile/01346397854558520528noreply@blogger.comBlogger5125tag:blogger.com,1999:blog-826910957631479390.post-39025139265867848702014-08-22T08:06:40.389-07:002014-08-22T08:06:40.389-07:00now that you mentioned it i remember that i used b...now that you mentioned it i remember that i used bdr_init_copy, i did this setup at 16 of july so i probably missed that in my notes, thanks for the comment, i will correct that. <br />(also thanks for reading)Evol Monkeyhttps://www.blogger.com/profile/01346397854558520528noreply@blogger.comtag:blogger.com,1999:blog-826910957631479390.post-77953357808073010752014-08-22T08:00:53.085-07:002014-08-22T08:00:53.085-07:00Also, note that you don't need to use two VMs....Also, note that you don't need to use two VMs. Per the BDR quickstart guide, a single node with servers running on two different ports is perfectly fine for testing purposes.<br /><br />Craig Ringerhttps://www.blogger.com/profile/02343803844223399065noreply@blogger.comtag:blogger.com,1999:blog-826910957631479390.post-38728660191330142522014-08-22T07:59:46.229-07:002014-08-22T07:59:46.229-07:00We released bdr/0.7 yesterday, so you might want t...We released bdr/0.7 yesterday, so you might want to try that. Lots of fixes and improvements. See the tag commit ("git show bdr/0.7") for a summary.<br /><br />It includes a number of fixes and improvements for global sequences and distributed DDL locking in particular.<br /><br />BTW, as guides like this will become outdated rapidly at BDR's current pace of development, please prominently highlight the user guide (which links to the quickstart): https://wiki.postgresql.org/wiki/BDR_User_Guide<br /><br />Finally: Where you write:<br /><br /><br /> Made a copy with pg_basebackup to a second $PGDATA directory, made the corrections on the connection strings in postgresql.conf so the 2 clusters could connect with each other and started both postgres servers.<br /><br /><br />... that should not have worked unless you then ran bdr_init_copy on the copied cluster to change the system identifiers and make some other fixups after the basebackup. bdr/0.6 is a long time ago in my timeline, but I'm quite sure we were already doing system identifier checks then, so you should've got an error along the lines of "system identifiers must differ". Please re-test on bdr/0.7 and write a report to pgsql-general with "BDR" prominently in the subject if you can still reproduce it.Craig Ringerhttps://www.blogger.com/profile/02343803844223399065noreply@blogger.comtag:blogger.com,1999:blog-826910957631479390.post-60304866199029184202014-08-22T07:36:14.525-07:002014-08-22T07:36:14.525-07:00Forgot the source: https://wiki.postgresql.org/wik...Forgot the source: https://wiki.postgresql.org/wiki/BDR_Conflicts#Distributed_Locking<br />Martin Gudmundssonhttps://www.blogger.com/profile/18173244973516224246noreply@blogger.comtag:blogger.com,1999:blog-826910957631479390.post-71388012005331688992014-08-22T07:25:06.038-07:002014-08-22T07:25:06.038-07:00Great article.
I also saw the same kind of issues...Great article. <br />I also saw the same kind of issues with locking as you didand my understanding is that only DDL has global locks, not DML. <br />Martin Gudmundssonhttps://www.blogger.com/profile/18173244973516224246noreply@blogger.com