I'm going to watch this carefully. We probably can't afford a server upgrade for awhile, and we're maxed out on our server at 4GB right now. If ES is going to exhaust our limited memory, we may have to pay someone to do an update to the Sphinx add-on instead. Or just tough it out and stick with vB 3.7 (which I have grown to really dislike lately).
I'm on vB 3.8. When I do eventually move over to XF, I'm going to try running ElasticSearch on a cheap Amazon EC2 instance so it won't affect the memory usage of my primary server. The "Small", standard instances (1.7GB RAM, 160GB HD) can be reserved for $97 per year.
Testmigration Posts Indexed: ~22.5 million Size on Disk: 19GB Total Min/Max Ram Allocated in config: 2gb/6gb 19GB..quite a lot!
1.1 million posts.... That command lists hundreds of items, but at the bottom is 2.5 Gig.....nodes/indices Speaking of that, the list still includes a ES instance that I installed (well, Slavik installed) on my test XF installation. I have removed that XF instance and even the db. Does ES eventually "forget" this? This was over a month ago and it still shows up in that du.
OK.....so I can delete it from the command line instead of with the ES delete index command? And everything else will then adjust itself as the ongoing indexes update? For clarity, there are two directories in "indices" - one for the older test setup and one for the existing live setup. I assume I simply delete the older one completely?? "Better to be safe than sorry", I say.......
Yes for example it would be something like /var/elasticsearch/*your es name*/nodes/0/indices In there you would have 2 folders, one for your old and one for your new. Just delete the old folder outright. It will have no effect on your new one.
Well, that got rid of 1/2 the index! Now: 1.3G total As far as server RAM, I think Linux eats up whatever you have as disk cache......so it may be harder to discern "free" RAM....... That is, if this is accurate.... http://www.linuxatemyram.com/
