Hi Earl, for the first error, it is not allowed for an index in SPARSE mode to have more than 5 rows per value. You have to choose another mode.
I let you read this blog post on the subject : http://www.doanduyhai.com/blog/?p=2058 I don't know if the following errors are related to the first one. Ludovic Le dim. 10 déc. 2017 à 03:42, Earl Lapus <earl.la...@gmail.com> a écrit : > Hi Jeff, > > On Sun, Dec 10, 2017 at 7:52 AM, Jeff Jirsa <jji...@apache.org> wrote: > >> That looks like a real problem, but it's not immediately clear to me what >> it is. >> >> It's possible that it's a genuine disk failure (do you see other signs of >> that disk having read/write failures?), or more likely, some race in SASI >> that is causing the index file to be removed while it's being updated. >> > > The disk is fine, so the SASI race issue is more likely the problem. I can > remove the SASI indices in the schema and see if those error logs will go > away. > > >> >> If possible, I encourage you to open a JIRA with as much info as you can >> (the stacks are great, sanitized logs would be good too). JIRA is here: >> https://issues.apache.org/jira/browse/CASSANDRA > > > I'll gather more logs and file a JIRA if I cannot find anything similar > that is reported already. > > >> >> >> You could also try upgrading to 3.11.1, there's always a chance it's been >> fixed (though I don't see anything in the 3.11.1 changelog that looks >> related). >> >> > I'll give this a shot and see what happens. > > Cheers, > Earl > >
