[Glass] Rc-Write-Write commit conflicts on Seaside session cache

[Johan Brichau]

Thanks for that info Dale.

They are occurring under load but I don't see too many of them.
So, I think everything is fine.

Johan

On 23 Sep 2013, at 16:52, "Dale K. Henrichs" <dale.henrichs at gemtalksystems.com> wrote:

> Johan,
> 
> If the commit is failing with a retryFailure, then it is possible that you just have too many sessions attempting to update the dictionary and the session with the retryFailure is just unlucky in that it hits the retry limit ... BTW, these retries are occuring down in the commit logic not at the http level ...
> 
> We have a rc lock that serializes rc retries, but the rc lock doesn't prevent other sessions from jumping in and attempting to do an update of the dictionary ... it may be necessary to look at alternative solutions form managing the cache ...
> 
> Dale
> ----- Original Message -----
> | From: "Johan Brichau" <johan at yesplan.be>
> | To: glass at lists.gemtalksystems.com
> | Sent: Friday, September 20, 2013 5:10:01 AM
> | Subject: [Glass] Rc-Write-Write commit conflicts on Seaside session cache
> | 
> | Hi,
> | 
> | I am seeing regular Rc-Write-Write conflict occurring on the Seaside
> | session cache (Seaside 3.0.8 / Glass 1-beta-9 / gemstone 2.4.4.1)
> | 
> | I noticed this should have been solved:
> | https://code.google.com/p/glassdb/issues/detail?id=80
> | But the config shows I am using the WARcLastAccessExpiryPolicy ...
> | 
> | So my first question: is there anything else I need to check to see
> | if it would not be the same bug?
> | 
> | Johan
> | _______________________________________________
> | Glass mailing list
> | Glass at lists.gemtalksystems.com
> | http://lists.gemtalksystems.com/mailman/listinfo/glass
> |