Implementing updates to source code executing on a plurality of compute nodes

Patent metadata

Field	Value
Publication number	US-9720676-B2
Application number	US-201213369879-A
Country	US
Kind code	B2
Filing date	Feb 9, 2012
Priority date	Feb 9, 2012
Publication date	Aug 1, 2017
Grant date	Aug 1, 2017

Abstract

Methods, apparatuses, and computer program products for implementing updates to source code executing on a plurality of compute nodes are provided. Embodiments include receiving, by a compute node, a broadcast update-notification message indicating there is an update to the source code executing on the plurality of compute nodes; in response to receiving the update-notification message, implementing a distributed barrier; based on the distributed barrier, halting execution of the source code at a particular location within the source code; based on the distributed barrier, updating in-place the source code including retaining workpiece data in memory of the compute node, the workpiece data corresponding to the execution of the source code; and based on completion of the updating of the source code, resuming with the retained workpiece data execution of the source code at the particular location within the source code where execution was halted.

First claim

What is claimed is:

1. An apparatus for implementing updates to source code executing on a plurality of compute nodes, the apparatus comprising a computer processor and a computer memory operatively coupled to the computer processor, the computer memory having disposed within it computer program instructions that when executed by the computer processor cause the apparatus to carry out the steps of:
receiving from a source code updater, by a compute node, a broadcast update-notification message indicating there is an update to the source code executing on the plurality of compute nodes;
in response to receiving the update-notification message, implementing, by the computer node, a distributed barrier on the compute node, the distributed barrier controlling implementation of updates to the source code, wherein the distributed barrier comprises instructions for the plurality compute nodes to gather at a particular step in the implementation of updates to the source code;
based on the distributed barrier, halting, by the compute node, execution of the source code at a particular location within the source code, wherein all of the plurality of compute nodes halt execution of the source code based on the distributed barrier, wherein the implemented distributed barrier prevents the compute node from resuming execution of the source code until each compute node in the plurality of compute nodes has completed updating the source code;
receiving by the compute node from the source code updater an update-now message that indicates each of the compute nodes in the plurality of compute nodes are ready to update the source code;
based on the distributed barrier, updating in-place, by the compute node, the source code including retaining workpiece data in memory of the compute node, the workpiece data corresponding to the execution of the source code;
wherein the workpiece data include state of variables, pointers, indexes, and other execution data generated during execution of the source code;
in response to completing the updating of the source code, updating, by the compute node, an indication of a source code in-use version to register that the compute node completed updating of the source code by sending to the source code updater a message to change a version of source code corresponding to the compute node;
receiving by the compute node from the source code updater an update-complete message indicating that the source code in-use version of each of the compute nodes of the plurality of compute nodes corresponds with a version of the updated source code; and
in response to receiving the update-complete message, resuming with the retained workpiece data, by the compute node, execution of the source code at the particular location within the source code where execution was halted, wherein upon resuming, all of the plurality of compute nodes resume execution of the source code.

2. The apparatus of claim 1 wherein the implemented distributed barrier prevents the compute node from updating the source code until each compute node of the plurality of compute nodes have indicated a readiness to update the source code.

3. The apparatus of claim 2 further comprising:
in response to receiving the update-notification message, determining, by the compute node, if the compute node is ready to update the source code;
if the compute node is ready to update the source code, transmitting, by the compute node, a ready-update message;
wherein updating in-place the source code is in response to receiving the update-now message; and
wherein resuming execution of the source code is in response to completing, by the compute node, the updating of the source code at the compute node.

4. The apparatus of claim 3 , wherein the update-notification message is an active message that includes an all reduce operation for gathering ready-update messages from each of the plurality of compute nodes.

5. A computer program product for implementing updates to source code executing on a plurality of compute nodes, the computer program product disposed upon a computer readable medium, wherein the computer readable medium is not a signal, the computer program product comprising computer program instructions that when executed by a computer cause the computer to carry out the steps of:
receiving from a source code updater, by a compute node, a broadcast update-notification message indicating there is an update to the source code executing on the plurality of compute nodes;
in response to receiving the update-notification message, implementing, by the computer node, a distributed barrier on the compute node, the distributed barrier controlling implementation of updates to the source code, wherein the distributed barrier comprises instructions for the plurality compute nodes to gather at a particular step in the implementation of updates to the source code;
based on the distributed barrier, halting, by the compute node, execution of the source code at a particular location within the source code, wherein all of the plurality of compute nodes halt execution of the source code based on the distributed barrier, wherein the implemented distributed barrier prevents the compute node from resuming execution of the source code until each compute node in the plurality of compute nodes has completed updating the source code;
receiving by the compute node from the source code updater an update-now message that indicates each of the compute nodes in the plurality of compute nodes are ready to update the source code;
based on the distributed barrier, updating in-place, by the compute node, the source code including retaining workpiece data in memory of the compute node, the workpiece data corresponding to the execution of the source code;
wherein the workpiece data include state of variables, pointers, indexes, and other execution data generated during execution of the source code;
in response to completing the updating of the source code, updating, by the compute node, an indication of a source code in-use version to register that the compute node completed updating of the source code by sending to the source code updater a message to change a version of source code corresponding to the compute node;
receiving by the compute node from the source code updater an update-complete message indicating that the source code in-use version of each of the compute nodes of the plurality of compute nodes corresponds with a version of the updated source code; and
in response to receiving the update-complete message, resuming with the retained workpiece data, by the compute node, execution of the source code at the particular location within the source code where execution was halted, wherein upon resuming, all of the plurality of compute nodes resume execution of the source code.

6. The computer program product of claim 5 wherein the implemented distributed barrier
prevents the compute node from updating the source code until each compute node of the plurality of compute nodes have indicated a readiness to update the source code.

7. The computer program product of claim 6 further comprising:
in response to receiving the update-notification message, determining, by the compute node, if the compute node is ready to update the source code;
if the compute node is ready to update the source code, transmitting, by the compute node, a ready-update message;
wherein updating in-place the source code is in response to receiving the update-now message; and
wherein resuming execution of the source code is in response to completing, by the compute node, the updating of the source code at the compute node.

8. The computer program product of claim 5 , wherein the update-notification message is an active message that inclu

Assignees

• Archer Charles J
• Carey James E
• Weaver Matthew R
• IBM

Inventors

• Archer Charles J
• Carey James E
• Weaver Matthew R

Classifications

• G06F9/522 Primary

  Barrier synchronisation · CPC title

• G06F8/67

  Physics · mapped topic

• G06F8/65 Primary

  Updates (security arrangements therefor G06F21/57) · CPC title

• G06F9/542

  Event management; Broadcasting; Multicasting; Notifications · CPC title

• G06F8/656

  while running · CPC title