Mongoose.Service Apparent Memory Leak with SoapEwsRestProvider

[SteveGregory]

Hi, 

I am seeing the Mongoose.Service memory usage climb up to 31gb over a period of about 8 hours, before the service stops. The only thing configured in the SmartConnector is two SoapEwsRestProvider endpoints. There are two REST clients reading data from the endpoints which are pointing to one enterprise server.

The SmartConnector version is 2.5.4 and EBO version 3.0.3.11

The attachment shows the last messages in the log before the service shuts down.

[ard]

Hi Steve,

Has SmartConnector been updated from a previous version? There's an issue where memory usage will grow if the Subscription/Notification tables in the database are very large when updated from 2.4.* to 2.5.*

--Armend

[SteveGregory]

Hi, I had the issue an earlier version of 2.4, so I updated to the latest version of 2.4.40 and had the same issue. I then updated to 2.5.4.18 and changed the endpoints from SBO REST providers to SOAP REST providers, relicensed the SmartConnector and had to get the clients to change all their log ids, and am having the same issue.

What is the fix for the issue you mentioned? Is it to clear the tables in the SQL DB? Can you please advise? Thanks

[SteveGregory]

Further to my previous message, i can confirm that the tables are mostly empty as we are only using endpoints. see screen shot attached. 

Originally, when i created the second endpoint, i duplicated the existing endpoint and I noticed that the two endpoints had the same 'Cache Tenant ID' which appeared to be causing a lot of errors (refer to 2022-08-26 crash log). I deleted one of the endpoints and instead created it from scratch. Those errors are gone now, but I am still getting the periodic issue, whereby the memory max's out and the service stops refer to 2022-08-27 crash log).

Also, the In memory cache has a lot of values, see screen shot. Could that be the issue?

Steve

[ard]

Looks like a different issue then. The fix for the issue I mentioned would be to truncate the tables related to notifications and subscriptions.

The logs provide very little info and the cache usage could be reasonable depending on your usage. I suggest you contact the SC team, wait for the process to grow to a couple of gigabytes and provide them with a memory dump, I think that would be easiest 🙂

--Armend

[SteveGregory]

Hi Armend,

Thank you for your reply. Can you tell me who are the SC Team? Do I just use the regular product support channel for this or do you a direct contact I can reach out to?

Steve

[CraigEl]

Yes Steve, standard product support process. Smart Connector framework is now supported by EBO R&D