I have raised Jira ticket https://issues.apache.org/jira/browse/IGNITE-24634
to capture this issue, and a related issue with cache local peeking
reporting OffHeap presence after cold restart of grid.
Raymond.
On Tue, Feb 25, 2025 at 12:44 PM Raymond Wilson <raymond_wil...@trimble.com>
wrote:
> Now I am confused.
>
> I have configured caches correctly to enable statistics (both via
> configuration and via direct setting of the flag on an existing cache.
>
> When I log the results for cache statistics (number of Gets and number of
> off-help Gets), I notice two things:
>
> - The total number of off heap Gets is always zero (yet I have a lot of
> RAM configured as OffHeap storage).
> - The total number of Get is very small compared to my other metrics that
> are already counting the number of Gets made to the caches.
>
> Right now I would suggest there is a bug somewhere here and cache
> statistics are broken.
>
> Raymond.
>
>
>
>
>
>
>
> On Tue, Feb 25, 2025 at 10:45 AM Raymond Wilson <
> raymond_wil...@trimble.com> wrote:
>
>> It seems you can modify cache statistics enablement for existing caches
>> using the setter on the cache instance. I was setting this flag in the
>> configuration for creating a new cache on the GetOrCreateCache(0 call,
>> which of course is ignored if the cache already exists.
>>
>> I am still unsure why the peeking does not work. Given the use of the
>> cache Get() statistics most of my use case for using peeking is satisfied.
>> However, I am inspecting the content of the payload to derive some
>> additional internal metrics beyond just calls count (eg: payload size)
>> which are failing given the peeking oddness.
>>
>> On Mon, Feb 24, 2025 at 11:23 PM Raymond Wilson <
>> raymond_wil...@trimble.com> wrote:
>>
>>> FYI, I am using Ignite 2.16.
>>>
>>> I have also been playing with the Ignite metrics. This allows me to see
>>> the numbers of Get()'s for each cache, but I have a suspicion I can't
>>> enable it for existing caches.Is that correct? At least, when I enable it
>>> on an existing system I just get zeros back, but a new system appears to
>>> track the statistics as expected.
>>>
>>> Thanks,
>>> Raymond.
>>>
>>>
>>> On Fri, Feb 21, 2025 at 12:44 PM Raymond Wilson <
>>> raymond_wil...@trimble.com> wrote:
>>>
>>>> Apologies, Change
>>>>
>>>> { /* It was read from persistence */ }
>>>>
>>>> for
>>>>
>>>> {
>>>> /* It needs to be read from persistence */
>>>> result = _cache.Get(key);
>>>> }
>>>>
>>>> On Fri, Feb 21, 2025 at 12:39 PM Raymond Wilson <
>>>> raymond_wil...@trimble.com> wrote:
>>>>
>>>>> I am experimenting with some metrics to help us understand how often
>>>>> Cache.Get() operations are satisfied by returning a value already
>>>>> existing in memory, versus those that are satisfied by needing to read
>>>>> that
>>>>> content from the persistent store.
>>>>>
>>>>> The way I am attempting to do this is by using TryLocalPeek(), eg:
>>>>>
>>>>> if (_cache.TryLocalPeek(key, out var result,
>>>>> CachePeekMode.Onheap, CachePeekMode.Offheap))
>>>>> { /* It was already in memory */ }
>>>>> else
>>>>> { /* It was read from persistence */ }
>>>>>
>>>>> Oddly, it appears almost all calls are satisfied by the TryLocalPeek()
>>>>> call, even on a completely cold restarted system.
>>>>>
>>>>> Can anyone confirm if my expected behaviour here is correct (and there
>>>>> is a bug), or if I am 'doing it wrong' and there is a better way?
>>>>>
>>>>> Thanks,
>>>>> Raymond.
>>>>>
>>>>>
>>>>> --
>>>>> <http://www.trimble.com/>
>>>>> Raymond Wilson
>>>>> Trimble Distinguished Engineer, Civil Construction Software (CCS)
>>>>> 11 Birmingham Drive | Christchurch, New Zealand
>>>>> raymond_wil...@trimble.com
>>>>>
>>>>>
>>>>> <https://worksos.trimble.com/?utm_source=Trimble&utm_medium=emailsign&utm_campaign=Launch>
>>>>>
>>>>
>>>>
>>>> --
>>>> <http://www.trimble.com/>
>>>> Raymond Wilson
>>>> Trimble Distinguished Engineer, Civil Construction Software (CCS)
>>>> 11 Birmingham Drive | Christchurch, New Zealand
>>>> raymond_wil...@trimble.com
>>>>
>>>>
>>>> <https://worksos.trimble.com/?utm_source=Trimble&utm_medium=emailsign&utm_campaign=Launch>
>>>>
>>>
>>>
>>> --
>>> <http://www.trimble.com/>
>>> Raymond Wilson
>>> Trimble Distinguished Engineer, Civil Construction Software (CCS)
>>> 11 Birmingham Drive | Christchurch, New Zealand
>>> raymond_wil...@trimble.com
>>>
>>>
>>> <https://worksos.trimble.com/?utm_source=Trimble&utm_medium=emailsign&utm_campaign=Launch>
>>>
>>
>>
>> --
>> <http://www.trimble.com/>
>> Raymond Wilson
>> Trimble Distinguished Engineer, Civil Construction Software (CCS)
>> 11 Birmingham Drive | Christchurch, New Zealand
>> raymond_wil...@trimble.com
>>
>>
>> <https://worksos.trimble.com/?utm_source=Trimble&utm_medium=emailsign&utm_campaign=Launch>
>>
>
>
> --
> <http://www.trimble.com/>
> Raymond Wilson
> Trimble Distinguished Engineer, Civil Construction Software (CCS)
> 11 Birmingham Drive | Christchurch, New Zealand
> raymond_wil...@trimble.com
>
>
> <https://worksos.trimble.com/?utm_source=Trimble&utm_medium=emailsign&utm_campaign=Launch>
>
--
<http://www.trimble.com/>
Raymond Wilson
Trimble Distinguished Engineer, Civil Construction Software (CCS)
11 Birmingham Drive | Christchurch, New Zealand
raymond_wil...@trimble.com
<https://worksos.trimble.com/?utm_source=Trimble&utm_medium=emailsign&utm_campaign=Launch>
