Thank you for that Gabriel, I have immediately reviewed your PR since I am responsible. I have been mostly working with jdbc configurations which is why I was not aware of causing this performance issue in a file-based setup.

I would like to call out again to all geoserver developers that I am always available to do for reviews in a reasonable time frame, especially if it is code that I have written or am very familiar with. So please do not hesitate to assign me if github suggests me.

Also, if a regression of this kind becomes apparent only after review and merge, obviously report it asap, and you can definitely bring it to my attention too. I have never responded in any other way then to make it a priority to fix things I did wrong. I have always done my best to contribute in the best way possible, including fixing many bugs and adding test coverage to prevent regression, but mistakes do happen with anyone. I enjoy cooperating in a positive way with anyone willing to do so, like I have recently been doing with Steve on jdbcconfig. My only shortcoming in this community is that I don't always manage to follow everything that is going on, but if you contact me directly  I will always respond very rapidly and do everything to assist.

Kind Regards

Niels

On 28/04/2019 08:29, Gabriel Roldan wrote:
More on startup time, added the corresponding layers to those 40k shapefile stores, and geoserver was choking in loading the gwc tile layers from datadir/gwc-layers.

The following PR[1] takes loading the tile layers back down to ~1.5s instead of ~183s. With #3450, full start up down from 280s to 24s.

It is too late now, tomorrow/Monday I'll try to create a jira issue to the whole startup time thing.

Again, a PR review would be appreciated.

Cheers,
Gabriel.

[1] <https://github.com/geoserver/geoserver/pull/3451>


El sáb., 27 de abr. de 2019 a la(s) 20:59, Gabriel Roldan ([email protected] <mailto:[email protected]>) escribió:

    Hi group,

    I've just created this PR
    <https://github.com/geoserver/geoserver/pull/3450>, which speeds
    up GeoServer initialization when there are lots of datastore info
    objects to load.

    I've a data directory with about 40K shapefile datastores (no
    layers, in order to check this particular patch) and startup time
    on my machine goes from 38s to 13s.

    PR review most appreciated.

    Cheers,
-- Gabriel Roldán



--
Gabriel Roldán


_______________________________________________
Geoserver-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/geoserver-devel
_______________________________________________
Geoserver-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/geoserver-devel

Reply via email to