Accelerating system dump capturing
Embodiments include accelerating capture of a system dump at system failure in a computer system. A method includes capturing a system snapshot that includes data from real memory frames of the computer system and storing the system snapshot in a storage. The method also includes monitoring periodically a change between the data in the system snapshot and a current state of the real memory frames of the computer system and updating the system snapshot according to the changed real memory frames based on a rate of change of the real memory frames. Further, in response to encountering the system failure, the snapshot is updated based on a delta between the system snapshot and the current state of the real memory frames. The snapshot on the storage is used as the system dump. The present document further describes examples of other aspects such as systems, computer products.
1. A method for accelerating capturing a system dump at system failure in a computer system, the method comprising:
capturing, by a snapshot accelerator, a system snapshot that comprises data from real memory frames of the computer system;
storing, by the snapshot accelerator, the system snapshot in an external storage;
monitoring periodically, by the snapshot accelerator, for a change between the data in the system snapshot and a current state of the real memory frames of the computer system;
updating, by the snapshot accelerator, the system snapshot on the external storage according to the changed real memory frames based on a rate of change of the real memory frames; and
in response to encountering the system failure:
determining, by the snapshot accelerator, a change between the data in the system snapshot and the current state of the real memory frames of the computer system;
updating by the snapshot accelerator the snapshot on the external storage according to the changed real memory frames; and
using the snapshot on the external storage as the system dump.
2. The method of claim 1 , wherein the snapshot accelerator is a coprocessor in the computer system, distinct from a central processing unit of the computer system.
3. The method of claim 1 , wherein the change between the data in the system snapshot and the current state of the real memory frames of the computer system is determined based on a plurality of change bits respectively associated with the real memory frames.
4. The method of claim 1 , further comprising appending, by the snapshot accelerator, data from registers of the computer system to the snapshot in response to the system failure.
5. The method of claim 1 , further comprising transmitting, by the snapshot accelerator, the snapshot for diagnosing the system failure.
6. The method of claim 1 , wherein the snapshot accelerator updates a portion of the system snapshot on the external storage in response to the rate of change of the real memory frames corresponding to the portion being below a predetermined threshold.
7. The method of claim 1 , wherein updating the system snapshot on the external storage comprises:
determining a rate of change of each portion from a plurality of portions in the changed data between the system snapshot on the external storage and the current state of the real memory frames;
ordering the portions in the changed data according to the rate of change; and
updating the system snapshot on the external storage according to the ordered portions.
8. The method of claim 1 , wherein updating the system snapshot on the external storage comprises updating portions of the changed data between the system snapshot on the external storage and the current state of the real memory frames according to a predetermined scheduling scheme.
9. The method of claim 1 , wherein the periodic monitoring is based on at least one of a predetermined time duration between successive comparisons of the system snapshot and the current state of the real memory frames of the computer system, and completion of a predetermined number of instructions between successive comparisons of the system snapshot and the current state of the real memory frames of the computer system.