random crashes... help needed (Pharo, Magma, Seaside... I dunno)

> Hi,
> I'm having random crashes in a production image. This crashes occurs without leaving any information available... it just crash. No PharoDebug.log, no dump, nothing... finally, yesterday I get a PharoDebug.log (which I'm attaching) with some error information... but frankly I don't know if this is the reason the image is crashing or some other random error.
> Some other elements I'm seeing is:
>
> - Seaside sessions are in fact never removed unless I manually execute garbage collect.
> - WeakArray FinalizationDependents is bigger than usual (90 instances),  elements are mostly WeakIdentityDictionary and MaWeakIdentityDictionary (most of them empty but some other with sizes from 700 to 4000).
>
> I'm using a Pharo 1.1.1 with Seaside 3.0 and Magma 1.1r2
> The VM is any VM (exuperi 0.15f or unix cog) .
> OS is "some kind of linux"... looks like a "custom suse".
>

> On 27 December 2010 23:49, Esteban Lorenzano <[hidden email]> wrote:
>> Hi,
>> I'm having random crashes in a production image. This crashes occurs without leaving any information available... it just crash. No PharoDebug.log, no dump, nothing... finally, yesterday I get a PharoDebug.log (which I'm attaching) with some error information... but frankly I don't know if this is the reason the image is crashing or some other random error.
>> Some other elements I'm seeing is:
>>
>> - Seaside sessions are in fact never removed unless I manually execute garbage collect.
>> - WeakArray FinalizationDependents is bigger than usual (90 instances),  elements are mostly WeakIdentityDictionary and MaWeakIdentityDictionary (most of them empty but some other with sizes from 700 to 4000).
>>
>> I'm using a Pharo 1.1.1 with Seaside 3.0 and Magma 1.1r2
>> The VM is any VM (exuperi 0.15f or unix cog) .
>> OS is "some kind of linux"... looks like a "custom suse".
>>
> Okay, you're using oscog.35 which predates new finalization..
> at least it is not in a play.
>
>
>
>> Well... and I'm completely lost. Any help will be very appreciated... even on ways on how to get more info of my crashing system (redirecting standard and error output does not works either)... And the access I have to the O.S. is also limited.
>>
>
> well, all i can say that you should start digging from here:
>
> 'Error: highBit is not defined for negative integers'
>
> what if some error message prints numbers and using highbit for on
> negative one? it will then loop forever,
> recursively until there is a no free space left.
> but that of course could be anything else.
>
>
>> Cheers,
>> Esteban
>>
>>
>
>
>
> --
> Best regards,
> Igor Stasenko AKA sig.
>

> Hi,
> I'm having random crashes in a production image. This crashes occurs without leaving any information available... it just crash. No PharoDebug.log, no dump, nothing... finally, yesterday I get a PharoDebug.log (which I'm attaching) with some error information... but frankly I don't know if this is the reason the image is crashing or some other random error.
> Some other elements I'm seeing is:
>
> - Seaside sessions are in fact never removed unless I manually execute garbage collect.
> - WeakArray FinalizationDependents is bigger than usual (90 instances),  elements are mostly WeakIdentityDictionary and MaWeakIdentityDictionary (most of them empty but some other with sizes from 700 to 4000).
>
> I'm using a Pharo 1.1.1 with Seaside 3.0 and Magma 1.1r2
> The VM is any VM (exuperi 0.15f or unix cog) .
> OS is "some kind of linux"... looks like a "custom suse".
>
> Well... and I'm completely lost. Any help will be very appreciated... even on ways on how to get more info of my crashing system (redirecting standard and error output does not works either)... And the access I have to the O.S. is also limited.
>
> Cheers,
> Esteban
>
>

> Hi,
> No, I shouldn't have any of those problems... but I get another PharoDebug.log with some more information... something sorting things on WeakArray.
> Well, I'm attaching the log, maybe this is more helpful than the other one.
>
> Esteban
>
> <PharoDebug.log>
>
> El 28/12/2010, a las 4:40p.m., Chris Muller escribió:
>
>> Hi Esteban, there are a couple of Magma-related items you might want to check:
>>
>> - Is it possible a proxy is being passed as an argument to a
>> primitive?  I've only ever seen this problem manifest as a primitive
>> failure, but maybe with the right primitive it could cause what you're
>> seeing..?  If it is, send #yourself to the argument in the primitive
>> sender.
>>
>> - Do you have BlockContexts or BlockClosures in your persistent
>> model?  Check your repository definition's 'classDefinitions'
>> dictionary, if they're in there, remember that only simple blocks are
>> supported, e.g., only reference the block args, like in a sortBlock.
>>
>>
>>
>> On Mon, Dec 27, 2010 at 4:49 PM, Esteban Lorenzano <[hidden email]> wrote:
>>> Hi,
>>> I'm having random crashes in a production image. This crashes occurs without leaving any information available... it just crash. No PharoDebug.log, no dump, nothing... finally, yesterday I get a PharoDebug.log (which I'm attaching) with some error information... but frankly I don't know if this is the reason the image is crashing or some other random error.
>>> Some other elements I'm seeing is:
>>>
>>> - Seaside sessions are in fact never removed unless I manually execute garbage collect.
>>> - WeakArray FinalizationDependents is bigger than usual (90 instances),  elements are mostly WeakIdentityDictionary and MaWeakIdentityDictionary (most of them empty but some other with sizes from 700 to 4000).
>>>
>>> I'm using a Pharo 1.1.1 with Seaside 3.0 and Magma 1.1r2
>>> The VM is any VM (exuperi 0.15f or unix cog) .
>>> OS is "some kind of linux"... looks like a "custom suse".
>>>
>>> Well... and I'm completely lost. Any help will be very appreciated... even on ways on how to get more info of my crashing system (redirecting standard and error output does not works either)... And the access I have to the O.S. is also limited.
>>>
>>> Cheers,
>>> Esteban
>>>
>>>
>>
>