Обсуждение: xlogreader v3/xlogdump v2
Hi everyone, At http://git.postgresql.org/gitweb/?p=users/andresfreund/postgres.git;a=shortlog;h=refs/heads/xlogreader_v3 git://git.postgresql.org/git/users/andresfreund/postgres.git you can find my attempt trying to bring the xlogreader from Heikki, as modified by Alvaro, into a state where it has the capabilities to be usable for BDR. This is *preliminary* work, to see whether people roughly agree with the API, there is some smoothing of edges left. Changes I made: * Add XLogFindNextRecord, to find the next valid xlog record >= an recptr * Move the page validation handling into xlogreader * Add support for reading pages which are only partially valid * Add callback as a replacement for emode_for_corrupt_record I don't like the last part, it seems ugly to me, but moving the full error processing/formatting to a callback seems to involve more work. I am willing to do that work, but would like some input first. The xlogdump utility itself is in a mostly good state, some parts of support infrastructure (ereport wrapper, realpathbackend, timestamptz_to_str, pfree) need some work. Any opinions? I chose not to send the patch but rely on the git repository above, its already somewhat large. If people prefer to see it fully on the list I am happy to oblige. Greetings, Andres Freund --Andres Freund http://www.2ndQuadrant.com/PostgreSQL Development, 24x7 Support, Training & Services
Hi, On 2012-12-04 18:52:13 +0100, Andres Freund wrote: > At > http://git.postgresql.org/gitweb/?p=users/andresfreund/postgres.git;a=shortlog;h=refs/heads/xlogreader_v3 > git://git.postgresql.org/git/users/andresfreund/postgres.git > you can find my attempt trying to bring the xlogreader from Heikki, as > modified by Alvaro, into a state where it has the capabilities to be > usable for BDR. > > This is *preliminary* work, to see whether people roughly agree with the > API, there is some smoothing of edges left. > > Changes I made: > * Add XLogFindNextRecord, to find the next valid xlog record >= an recptr > * Move the page validation handling into xlogreader > * Add support for reading pages which are only partially valid > * Add callback as a replacement for emode_for_corrupt_record > > I don't like the last part, it seems ugly to me, but moving the full > error processing/formatting to a callback seems to involve more work. I > am willing to do that work, but would like some input first. > > The xlogdump utility itself is in a mostly good state, some parts of > support infrastructure (ereport wrapper, realpathbackend, > timestamptz_to_str, pfree) need some work. I pushed a new version of the patch with some fixes, more comments and a slightly changed read_page callback API. Greetings, Andres Freund --Andres Freund http://www.2ndQuadrant.com/PostgreSQL Development, 24x7 Support, Training & Services