Hi Mikhail, Do you have logs for when this happens? Maybe we could create a Jira to track it?
On Thu, Dec 14, 2017 at 4:12 AM, Mikhail Khludnev <m...@apache.org> wrote: > Hello, Devs. > > I observe interesting restore failure on SolrCloud 6.6. > When the second replica (1) is added, it recovers from the first one (0), > which is expected to be restored already. It usually fine, but quite > regularly it recovers from empty replica, like it happens before restore is > done. I briefly checked overseer log it expose enough level of sanity: > adding the second happens after the first one's recovery. > > Have you seen such sort of race-condition? > > -- > Sincerely yours > Mikhail Khludnev >
