[Mediawiki-l] updating wiki internal data after importDump
dnessett at yahoo.com
Tue Jan 26 21:44:23 UTC 2010
I have imported a daily dump of 121,000 pages from a wiki running 1.13.2. The importDump took 80 hours and when I ran rebuildAll, the resulting job queue was about 250,000 jobs. It took 10.5 hours to process them. After doing this, I downloaded a subsequent daily dump (from 6 days subsequent to the original daily dump) and imported it. That took about 30 minutes.
Now I want to update the internal data structures so my personal copy of the wiki is complete. I decided not to use rebuildAll again, since the daily dump has no images. Also, the wiki is postgres based, so rebuildtextindex doesn't work (it is only for MySQL). So, I thought I would use refreshLinks. However, that is running almost as slowly as the original importDump. After processing 2200 pages and interpolating forward, I estimate it will take about 40 hours to complete.
I have several questions:
+ Are there any settings that might make both importDump and refreshLinks do their work more quickly?
+ Is it necessary to run refreshLinks after an incremental importDump?
+ Are there other maintenance scripts I should run?
More information about the MediaWiki-l