"reason [do not exists on master, act as master failure" um
-
-
-
@bobpoekert LMAO if your convergence times are 300+ seconds. Unclear if this cluster is ever coming back.0 retweets 0 favorites -
@aphyr when I hit that scenario what fixed it was restarting the node that that node thought was master0 retweets 0 favorites -
@bobpoekert also I'm growing increasingly convinced that ES is not doing sync replication by default.0 retweets 1 favorite -
@aphyr me too, with my eskka discovery plugin i'm not seeing multiple masters afaict, yet still fail jepsen with 1-2 lost writes0 retweets 0 favorites -
@shikhrr Yeah. I've got a whole series of test cases leading to lost data or hung clusters. Random partitions, simple partitions, single-0 retweets 0 favorites -
@shikhrr node repeated partitions, even a single node being isolated once is enough to cause write loss AFAICT.0 retweets 0 favorites
@aphyr btw, "when a quorum of two is one" https://github.com/elasticsearch/elasticsearch/issues/6482 …
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
Arf Installation
bob
shikhar bhushan