On 2013-10-10 10:54 PM, "Klaus Berkling" <webobje...@berkling.us> wrote:
>On Oct 10, 2013, at 8:57 AM, Chuck Hill <ch...@global-village.net> wrote: > >> Do any of your requests take a long time to process (lots of database >> access or calculations)? Wotaskd will send the requests to another >> instance if one is too slow in responding. For most applications this >>is >> a useless attempt at recovery. You can try increasing the Connect and >> Receive Timeout in JavaMonitor under Adaptor Settings for this >> application. If that fixes the problem then you have requests that need >> to be optimized or done in a background thread. >> >> Chuck >> > >I'm pretty sure you are seeing your instance not responding and so some >other instance is getting the request. >Works only well for direct actions :-) Session-less direct actions at that. Unless you are using Ramsey's distributed sessions. In that case, component actions work too. > >In addition to what Chuck said, check the adaptor threads and listen >queue size in the Application settings. Typically I used min 10, max 20, >queue 20. This is very app specific. It's better to let connections queue >up than to let them go to a different instance. Well, up to a point. Beyond that the user's think there is a problem and start clicking again or reloading the page. Then your app gets overloaded and no one gets any service. It is very app specific, but I would usually advocate lower numbers. Your app, I know, is "special". ;-) Chuck >Also check the retries and connection pool size in the Load Balancing and >Adaptor Settings. Both should be greater then 0, less then 10. >Probably superfluous but these numbers should not be really small and not >really large. > >You might be able to 'expose' this problem better by running only one >instance, you might get more informative errors. > > >> >> >> >> >> On 2013-10-10 1:16 AM, "Alexander Spohr" <a...@freeport.de> wrote: >> >>> >>> Am 10.10.2013 um 09:42 schrieb Schoenenberger Dominique >>> <dominique.schoenenber...@nagra.com>: >>> >>>> It's like the adaptor (?) gives the request to the wrong instance. >>> >>> Yes. Instance 12 may be offline or not reachable (responds too slow) >>> >>> >>> _______________________________________________ >>> Do not post admin requests to the list. They will be ignored. >>> Webobjects-dev mailing list (Webobjects-dev@lists.apple.com) >>> Help/Unsubscribe/Update your Subscription: >>> >>>https://lists.apple.com/mailman/options/webobjects-dev/chill%40global-vi >>>ll >>> age.net >>> >>> This email sent to ch...@global-village.net >> >> >> _______________________________________________ >> Do not post admin requests to the list. They will be ignored. >> Webobjects-dev mailing list (Webobjects-dev@lists.apple.com) >> Help/Unsubscribe/Update your Subscription: >> >>https://lists.apple.com/mailman/options/webobjects-dev/webobjects%40berkl >>ing.us >> >> This email sent to webobje...@berkling.us > > > >kib > >Klaus Berkling >www.berkling.us | @kiberkli > > > > > _______________________________________________ >Do not post admin requests to the list. They will be ignored. >Webobjects-dev mailing list (Webobjects-dev@lists.apple.com) >Help/Unsubscribe/Update your Subscription: >https://lists.apple.com/mailman/options/webobjects-dev/chill%40global-vill >age.net > >This email sent to ch...@global-village.net _______________________________________________ Do not post admin requests to the list. They will be ignored. Webobjects-dev mailing list (Webobjects-dev@lists.apple.com) Help/Unsubscribe/Update your Subscription: https://lists.apple.com/mailman/options/webobjects-dev/archive%40mail-archive.com This email sent to arch...@mail-archive.com
