Open ceefour opened 10 years ago
We are trying to reproduce this issue. Can you provide content of your yago.gridgain.xml
configuration file? You can use http://pastebin.com/ to paste your code.
Sure, here's it: https://github.com/ceefour/lumen-kb/blob/e50b8fc0690c078fcec1c037be9496f1dc10937f/cli/src/main/resources/id/ac/itb/ee/lskk/lumen/yago/yago.gridgain.xml
If I can find reproducible steps I'll let you know.
This is a pretty rare bug which is reproduced when using full-scan query on changing topology. It was fixed and will be available in upcoming 6.2.0 release. As a workaround, you can now enable NEAR_PARTITIONED distribution mode on all nodes, or start your 'client' node that initiates query in CLIENT_ONLY or NEAR_ONLY distribution mode.
Thank you! Wow, you're good...
Java8 code: (https://github.com/ceefour/lumen-kb/blob/e50b8fc0690c078fcec1c037be9496f1dc10937f/cli/src/main/java/id/ac/itb/ee/lskk/lumen/yago/YagoLabelLookupCli.java)
After a few edits and runs, GridGain throws the following error. This error never happens on the first invocation (i.e. after starting 2 nodes, then running the compute)