Can anybody please point me to a link, where I can find detailed description of index.recovery.* settings? Tried to google it, but found this:
All its telling is what are the defaults, nothing much. Also, I would appreciate if anyone can tell me the recommended settings in case of non-SSD (cylindrical) drives. Please help!
That would be the defaults.
What is the problem you have?
Hey @warkolm ,
Actually, I wanted to speed up the recovery process. Suppose, if one of my nodes goes down, and then comes up, it takes a lot of time to recover. Till the recovery is completed, I get weird Shard related exceptions, if I search for an index not recovered yet.
Could you please tell me what settings I should twiddle, so that I can speed up the recovery process?
Yeah. What's so special about it? All it have is a list of settings without any description. I myself am not good at documentation, but I think every single configuration setting deserves a paragraph explaining its functionality and impact, Do's and Don'ts, etc.
then i supposed with that settings, you should get an idea what can you change but subject to the hardware/software and the use case in your environment. there is not like definitive answer such that you should set cluster.routing.allocation.node_concurrent_recoveries to 10. this should be gradual fine tune process to reach a level that is desirable. you can reference here https://www.loggly.com/blog/nine-tips-configuring-elasticsearch-for-high-performance/ tip number8, to see which parameter to change so you get an start idea what setting you can change to speed up recovery process. you most likely need to have monitoring in place since cluster recovery demand a lot of disk i/o and network i/o. this allow you to quickly gauge the difference before and after settings were changed.