Are you running the client on docker environment for a real cluster ? Can you list the output of 'uname -a' from one of the cluster nodes ?
Is it possible to narrow the scenario in a unit test ? For the row you're writing to, locate the region hosting it. On the region server for this region, was there active connection from the client ? Thanks On Fri, Sep 29, 2017 at 8:45 AM, Andrzej <[email protected]> wrote: > W dniu 29.09.2017 o 17:14, Ted Yu pisze: > >> From client side, you can specify any column you want to write. >>> >> If the column happens to exist for the row, new value would be written. >> >> For the code snippet, there is getTable() and Put(). >> Can you separate into two statements to narrow down which call hangs ? >> >> Which commit are you using in the branch ? >> >> Did you build with buck ? >> > > I build with buck, I use commit e2bb10a8708e7a675558ca0aad9e340823573ff6 > from 2017-09-22 > > messages: > getTable hTable = 0x10000 > getTable hEnv=1 hTable & 0xffff=0 > getTable, after get() > after getTable, before Put, table=1800d300 > > code: > my_printf("hbPutColumns before put\n"); > Table *table = getTable(hTable); > my_printf("after getTable, before Put, table=%x\n",table); > table->Put(*put); > my_printf("end hbPutColumns\n"); > > table is not null, it is raw pointer from unique pointer >
