A process called "omiengine" consumes too much memory

Question

_{Tuesday, October 16, 2018 9:23 AM}

Hi,

Theres a process called "omiengine", that takes way too much memory from the VM (almost half). The total memory amount is 3.5G, not much, but still, 50% memory for just monitoring and sending data to azure is way too much.

https://twitter.com/AzureSupport/status/1052118073283629056

How to resolve this issue?

Thanks in advance,

Jaan

PS!

Unfortunately I cant post an image or link here, because "

• Body text cannot contain images or links until we are able to verify your account.

"

You guy's should link this social account to the subscriber account, so I can be verifies, after all I am signing in from that account, seems like real nonsense why I need to verify again, you dont even send me a verification e-mail so that I could do that. If you don't have the competence to do a e-mail verification feature, Microsoft, then maybe I can help you.

All replies (3)

_{Tuesday, October 16, 2018 11:13 AM | 1 vote}

Hi,

This document might help with your troubleshooting. Howerver, for memory leaks it does advise to contact support as you already have done:

https://github.com/Microsoft/omi/blob/master/Unix/doc/diagnose-omi-problems.md

Diagnosing memory problems in OMI itself

OMI does have some diagnostic aids to assist in detection of memory leaks in OMI itself (these would be in either omiserveror omiengine processes). However, use of these aids requires not only a debug version of OMI, but engineering expertise in interpreting the results. Thus, if you believe you have diagnosed a memory leak on OMI itself (not in omiagent, but in omiserver or omiengine), you'll need to engage with Microsoft Support to help pin down the problem.

Do be aware that the OMI server/engine can grow based on load put upon it, and this growth is normal. A memory leak would thus be considered unbound memory growth such that OMI grows over time, without bound, based on time the OMI server was running and under load. Such growth can generally be seen over several days, regardless of load level.

_{Friday, October 19, 2018 7:29 AM}

Hope this answers your question, if so please mark as answered.

Thanks

_{Tuesday, November 20, 2018 5:35 PM}

Sorry, not helpful. The <g class="gr_ gr_66 gr-alert gr_spell gr_inline_cards gr_run_anim ContextualSpelling" data-gr-id="66" id="66">omiengine</g> process uses almost always about half of the memory, so in my opinion, it's still too much. Please consider this as an issue to the engineering team to optimize this.