Hi Andrew,
Thanks for that -- so far things are looking stable after making that
change.
-T
On 19/09/13 13:27, Andrew Stone wrote:
Hi Toby,
Can you try raising the pb_backlog to 128 in your riak app.config on
each node. It's likely those disconnect errors are left over from the
stampede of connections from the CS connection pool on startup. For one
reason or another the resets don't come through and the hanging
disconnected socket isn't discovered until the next send attempt.
-Andrew
On Wed, Sep 18, 2013 at 10:31 PM, Toby Corkindale
<toby.corkind...@strategicdata.com.au
<mailto:toby.corkind...@strategicdata.com.au>> wrote:
On 19/09/13 11:17, Luke Bakken wrote:
The "error: disconnected" message is a good clue. If you can provide
log files that may point to the cause.
See below for logs from riak and riak cs, for a roughly five minute
window during which I'd fiddled with the load balancer and DNS to
try and isolate this server from other requests, and only send mine
to it.
Unfortunately I really don't see much in there apart from the
disconnected messages. I did note that there were warnings about
system hitting high watermarks for memory at one point though.
That's a bit odd, as the machine wasn't near max capacity at any
point - it's only 8GB, but 'free' was reporting that most was being
used just for buffers and cache at the time; as follows:
ie.
total used free shared buffers
cached
Mem: 8176412 3803404 4373008 0 34112
3052340
-/+ buffers/cache: 716952 7459460
Swap: 3903484 0 3903484
_______________________________________________
riak-users mailing list
riak-users@lists.basho.com
http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com