Automated root cause determination using a pre-trained generative model
Systems and methods are provided for determining a root cause of an incident that occurred in a 5G/6G multi-access edge computing and core network system. In particular, the disclosed technology is directed to using a pre-trained generative model to determine the root cause for an incident as recorded in event data of a system log. The present disclosure generates a prompt for the generative model to determine the root cause for an incident as recorded in the system log. The prompt comprises a combination including event data from a system log, function hierarchy graph data, and function information as grounding information in a prefix of the prompt. The prompt further includes a question that requests determining a root cause of an incident as recorded in the event data. An answer as generated by the pre-trained generative model indicates of the root cause and a function call associated with the incident.
1 . A computer-implemented method, comprising:
retrieving a set of data including:
event data of a system log, wherein the event data indicates an incident,
function hierarchy graph data, and
function information;
generating a prompt for transmission to a pre-trained generative model, wherein the prompt includes grounding information and a question, the grounding information includes the event data, the function hierarchy graph data, and the function information, and the grounding information as a part of the prompt causes the pre-trained generative model to identify a root cause of an incident in a predetermined domain as described in the grounding information;
receiving, in response to the transmission of the prompt, a natural language text from the pre-trained generative model, wherein the natural language text includes a description of the root cause of the incident with accuracy by restricting the root cause in the predetermined domain according to the grounding information; and
displaying the description of the root cause.
2 . The computer-implemented method according to claim 1 , wherein the incident is associated with an error occurred in an edge server of a 5G multi-access edge computing network.
3 . The computer-implemented method according to claim 1 , wherein the function information includes fingerprint data associated with a function, the event data includes the fingerprint data, and the function hierarchy graph data includes the fingerprint data.
4 . The computer-implemented method according to claim 1 , wherein the prompt is based on the natural language text.
5 . The computer-implemented method according to claim 1 , further comprising:
retrieving a code associated with program instruction of an edge server in a 5G MEC system;
generating a summary prompt, wherein the summary prompt includes the code as the grounding information;
transmitting the summary prompt;
receiving a summary text associated with the code from the pre-trained generative model; and
displaying the summary text for insertion into the code.
6 . The computer-implemented method according to claim 1 , further comprising:
retrieving a code associated with program instruction of an edge server in a 5G MEC system;
generating a prompt for a missing function call, wherein the prompt for a missing function includes the function information and the code as the grounding information and the question, wherein the question includes a request for identifying a missing exception handler function call in the code;
transmitting the prompt for a missing function call;
receiving an answer for a missing function call from the pre-trained generative model, wherein the answer for a missing function call includes a textual descriptions of the missing exception handler function call in the code, and the textual descriptions of the missing exception handler function call includes fingerprint data associated with the missing exception handler function call; and
displaying a code snippet associated with the missing exception handler function call, wherein the code snippet includes the fingerprint data.
7 . The computer-implemented method according to claim 1 , wherein the function hierarchy graph data indicates a hierarchical caller-callee relationship between two function calls in program instruction.
8 . The computer-implemented method according to claim 1 , wherein the description of the root cause associated with the incident comprises:
fingerprint data identified in the event data,
a first textual description of a function call associated with the fingerprint data,
a second textual description of the incident according to the fingerprint data, and
a sample code of calling an exception handler function call for preventing the incident.
9 . The computer-implemented method according to claim 8 , further comprising:
retrieving a code associated with the fingerprint data;
updating the code according to the sample code; and
storing the code in a code repository database.
10 . A system for determining a root cause of an event using a pre-trained generative model, the system comprising:
a processor; and
a memory storing computer-executable instructions that when executed by the processor cause the system to execute operations comprising:
retrieving a set of data including:
event data of a system log, wherein the event data indicates an incident,
function hierarchy graph data, and
function information;
generating a prompt for transmission to the pre-trained generative model, wherein the prompt includes grounding information and a question, the grounding information includes the event data, the function hierarchy graph data, and the function information, and the grounding information as a part of the prompt causes the pre-trained generative model to identify a root cause of an incident in a predetermined domain as described in the grounding information;
receiving, in response to the transmission of the prompt, a natural language text from the pre-trained generative model, wherein the natural language text includes a description of the root cause of the incident with accuracy by restricting the root cause in the predetermined domain according to the grounding information; and
displaying the description of the root cause.
11 . The system according to claim 10 , wherein the incident is associated with an error occurred in an edge server of a 5G multi-access edge computing network.
12 . The system according to claim 10 , wherein the function information includes fingerprint data associated with a function, the event data includes the fingerprint data, and the function hierarchy graph data includes the fingerprint data.
13 . The system according to claim 10 , wherein the function hierarchy graph data indicates a hierarchical caller-callee relationship between two function calls in program instruction.
14 . The system according to claim 10 , wherein the description of the root cause associated with the incident comprises:
fingerprint data identified in the event data,
a first textual description of a function call associated with the fingerprint data,
a second textual description of the incident according to the fingerprint data, and
a sample code of calling an exception handler function call for preventing the incident.
15 . The system according to claim 14 , the processor further configured to cause the system to execute operations comprising:
retrieving a code associated with the fingerprint data;
updating the code according to the sample code; and
storing the code in a code repository database.
16 . A computer storage media storing computer-executable instructions that when executed by a processor cause a computer system to execute operations comprising:
retrieving a set of data including:
event data of a system log, wherein the event data indicates an incident,
function hierarchy graph data, and
function information;
generating a prompt for transmission to a pre-trained generative model, wherein the prompt includes grounding information and a question, the grounding information includes the event data, the function hierarchy graph data, and the function information, and the grounding information as a part of the prompt causes the pre-trained generative model to identify a root cause of an incident in a predetermined domain as described in the grounding information;
receiving, in response to the transmission of the prompt, a natural language text from the pre-trained generative model, wherein the natural language text includes a description of the root cause of the incident with accuracy by restricting the root cause in the predetermined domain according to the grounding information; and
displaying the description of the root cause.
17 . The computer storage media of claim 16 , wherein the incident is associated with an error occurred in an edge server of a 5G multi-access edge computing network.
18 . The computer storage media of claim 16 ,
wherein the function hierarchy graph data indicates a hierarchical caller-callee relationship between two function calls in program instruction, and
wherein the function information includes fingerprint data associated with a function, the event data includes the fingerprint data, and the function hierarchy graph data includes the fingerprint data.
19 . The computer storage media of claim 16 , wherein the description of the root cause associated with the incident comprises:
fingerprint data identified in the event data,
a first textual description of a function call associated with the fingerprint data,
a second textual description of the incident according to the fingerprint data, and
a sample code of calling an exception handler function call for preventing the incident.
20 . The computer storage media of claim 19 , the processor further causes a computer system to execute operations comprising:
retrieving a code associated with the fingerprint data;
updating the code according to the sample code for preventing the incident from occurring in a subsequent operation; and
storing the code in a code repository database.