Michael, I think I already recommended that ;)
Here our 2 clusters both run with that config. J-D On Mon, Jan 26, 2009 at 9:15 AM, Michael Dagaev <michael.dag...@gmail.com>wrote: > Thanks, Jean-Daniel. > Would you recommend setting dfs.datanode.socket.write.timeout=0 in > hadoop-site.xml ? > > On Mon, Jan 26, 2009 at 4:08 PM, Jean-Daniel Cryans <jdcry...@apache.org> > wrote: > > Michael, > > > > It seems there are many reasons that it can timeout, the example given in > > HADOOP-3831 is a slow reading client. > > > > J-D > > > > On Mon, Jan 26, 2009 at 8:46 AM, Michael Dagaev < > michael.dag...@gmail.com>wrote: > > > >> Jean-Daniel, > >> > >> Property dfs.datanode.socket.write.timeout is not set > >> in hadoop-site.xml. > >> It does not appear in hadoop-default.xml either. > >> > >> Do you know why the data node sockets timed out ? The host does not > >> look overloaded. > >> > >> Thank you for your cooperation, > >> M. > >> > >> On Mon, Jan 26, 2009 at 3:33 PM, Jean-Daniel Cryans < > jdcry...@apache.org> > >> wrote: > >> > Michael, > >> > > >> > You don't see anything in your region server logs? Mmm., we usually > get > >> > those if we don't set the following in the hadoop-site.xml file: > >> > > >> > <property> > >> > <name>dfs.datanode.socket.write.timeout</name> > >> > <value>0</value> > >> > </property> > >> > > >> > See if it stops the exception. In any case, until Hadoop 0.18.3 and > >> Hadoop > >> > 0.19.1, you should probably still use that config to be safe. > >> > > >> > J-D >
