No, the server never does recover from the leak. There was a time or two
where I saw the process itself drop down to using 45 MB of memory or so,
but the total commit charge in Task Manager was still saying 2031M/3938M
or something like that. But, for the most part, the server process
stays at a memory usage level of around 1GB.
Zac Spitzer wrote:
Does the server recover from the memory leak or do you need to restart
the server?
On Tue, Feb 26, 2008 at 7:37 AM, Jonathan Manafi
<[EMAIL PROTECTED]> wrote:
I have narrowed it down to 2 of our layers, where when they are
available, the ajax viewer seems to hang at small zooms, or there
becomes a horrendous memory leak, where the smaller the scale becomes,
the more memory is consumed by mgserver.exe. The most I witnessed turned
out to be around 1.2 GB of memory consumed. With these two layers off,
I can zoom down to 1:.001 with no problems. Both of these layers have
line styles, and one has a single label at a particular zoom level, but
they are both sparse compared to a couple of other layers that have
multiple line styles and/or labels. All of our data being applied is
.SDF files.
Traian Stanev wrote:
> Try identifying which layer causes the slow down, by selectively turning off
layers. Then see what's peculiar about the layer -- does it have a line style, or
perhaps labeling. There is a known issue around that (layers with lots of labels
at high zoom), which will be fixed very soon.
>
> Traian
>
>
>
>> -----Original Message-----
>> From: [EMAIL PROTECTED] [mailto:mapguide-users-
>> [EMAIL PROTECTED] On Behalf Of Jonathan Manafi
>> Sent: Monday, February 25, 2008 12:54 PM
>> To: MapGuide Users Mail List
>> Subject: Re: [mapguide-users] map refresh timeout
>>
>> As an update:
>>
>> We began testing our data upgrading to MGOS 2.0 RC4 without using the
>> fusion library, and we are noticing the same issues. The AGG renderer
>> actually causes mgserver.exe to lock up at small zoom scales; today it
>> happened at 1:254, whereas the GD renderer has yet to lock up, but it
>> still takes a good period of time to update the map extents. Compared
>> to
>> MGOS 1.2 with the same dataset, both renderers perform worse when
>> zoomed
>> to a small scale, where, otherwise, they take only split seconds to
>> update when zoomed far out.
>>
>> Has anyone seen this behavior with their data, as this could be a
>> necessary function for our clients?
>>
>>
>> J Manafi wrote:
>>
>>> I have experienced extremely long refresh rates when the zoom scale
>>>
>> is really
>>
>>> low (~1:33 or so). The time it takes to refresh for the Sheboygan
>>>
>> example is
>>
>>> at least 10x normal, and on a custom map, MapGuide server pretty much
>>>
>> hangs
>>
>>> at ~1:236 (currently sitting and waiting). Has anyone else noticed
>>>
>> this
>>
>>> behavior at low scales?
>>>
>>>
>> _______________________________________________
>> mapguide-users mailing list
>> mapguide-users@lists.osgeo.org
>> http://lists.osgeo.org/mailman/listinfo/mapguide-users
>>
_______________________________________________
mapguide-users mailing list
mapguide-users@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/mapguide-users
_______________________________________________
mapguide-users mailing list
mapguide-users@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/mapguide-users
