Does OneAgent or ActiveGates retain data when not able to connect upstream?

[dirk_loosen]

We are trying to decide on whether we should use gateways or not and our primary concern is monitoring data loss. I see a similar question was already asked but not answered. Does anyone know if OneAgent retains data for a while and how long if it loses the connection to a gate and also how long a gate may keep data if it loses connection to the cluster?

[franz_soengen · Accepted solution]

Hi Dirk,

OneAgents as well as ActiveGates have a limited caching functionality for short outages. The time that can cover depends on the amount of data your Agents generate though. My last information is about 10 MB on Agents and approx. 100MB on ActiveGates but there the data is compressed so it can store data for a longer time than 10 Agents could (edit: there is no memory limit for caching on the ActiveGate, there are time limits defined after which messages are treated as outdated and deleted from the cache, and in terms of memory there is no limit so ActiveGate will try to utilize all heap available according to needs).

So to sumarize it's advisable to use ActiveGates for this scenario but they also have a limit.

best regards
Franz

[dirk_loosen]

Thanks @Franz S.. Makes sense that one cannot give a time, but it gives me some indication of how long before we lose data. Much appreciated. Regards, Dirk

[franz_soengen]

Hi Dirk,

it's very hard to make an educated guess here because the amount of generated data varies greatly by technologies used on the machine, configuration and throughput. Some low traffic servers may run up to an hour on this 10MB while some high-load Application server may have to discard data after a few minutes.

best regards
Franz

[GilesDay]

Is there any official documentation on this?

[Erik_Ostermuell]

Could you please detail the circumstances when OneAgents and ActiveGates stop retaining data?

Would like to know what control we have in the matter.

* Does adding RAM help?

* Must DT configuration also be tweaked? perhaps to increase buffer size or something?

* Are there special events/messages in log files, or other indicators of data loss?