We're working on profiling from the DB side. It's difficult to get a debug GeoServer setup against the environment showing the problem.
On Tue, Sep 1, 2015 at 2:47 PM, Jody Garnett <[email protected]> wrote: > So are you in position to profile the code? We recently enabled the > database online tests again, but they test conformance, not performance. > > -- > Jody Garnett > > On 1 September 2015 at 14:38, Martin Davis <[email protected]> wrote: > >> Yes, we are suspicious that the metadata queries are returning a lot of >> rows when no schema is specified. But can't confirm this is happening, >> until we can get DB-level tracing enabled. >> >> And as you say, why would this be happening on every GetMap ? And why >> happening in one environment and not in a similar different one? >> >> On Tue, Sep 1, 2015 at 2:13 PM, Jody Garnett <[email protected]> >> wrote: >> >>> I have a silly suggestion, when not using a schema is the data store >>> getting back an amazingly large number of oracle tables .. checking each >>> one for a spatial index and so on? >>> >>> I would expect that to take a bit longer on startup ... but you are >>> indicating that every GetMap request is consistently slow. >>> >>> -- >>> Jody Garnett >>> >>> On 1 September 2015 at 12:46, Martin Davis <[email protected]> wrote: >>> >>>> An update on this issue. >>>> >>>> As Andrea predicted, using a JNDI connection made no difference to the >>>> performance issue (no schema still substantially slower than using a >>>> schema). >>>> >>>> We're now attempting to do Oracle logging to try and see what's getting >>>> run that might slow a map request down. (We have only limited access to >>>> the box where the problem shows up). Results are not conclusive so far, >>>> but we think we are seeing several queries being run over as many as 5 >>>> different "Geoserver sessions" during a single map request. Cannot tell >>>> what these are yet, but seems likely they are metadata queries. This is >>>> odd, since we are not seeing this happen in another similar environment. >>>> One difference is that we are connecting via an Oracle Service rather than >>>> a SID in the slow environment. Would be odd if this was the cause, though. >>>> >>>> >>>> On Fri, Jul 31, 2015 at 10:02 AM, Martin Davis <[email protected]> >>>> wrote: >>>> >>>>> Thanks, Andrea. >>>>> >>>>> We're using GeoServer 2.6.0 >>>>> >>>>> The performance issue occurs for map requests - so wouldn't this be >>>>> something different to the issue with slow metadata loading (The metadata >>>>> retrieval is an issue we've seen as well, but it only hurts the admin, not >>>>> the users, so we're less caring about that 8^). >>>>> >>>>> We'll probably try using a JNDI pool and see whether that helps at >>>>> all. If so, we may just use that approach. If not, we'll be looking for >>>>> a >>>>> code fix - which we can likely get funded and contribute back. >>>>> >>>>> >>>>> >>>> >>>> ------------------------------------------------------------------------------ >>>> >>>> _______________________________________________ >>>> Geoserver-users mailing list >>>> [email protected] >>>> https://lists.sourceforge.net/lists/listinfo/geoserver-users >>>> >>>> >>> >> >
------------------------------------------------------------------------------
_______________________________________________ Geoserver-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/geoserver-users
