Distributed resource contention detection and handling

• US 9,870,369 B2
• Filed: 07/08/2016
• Issued: 01/16/2018
• Est. Priority Date: 05/05/2010
• Status: Active Grant

• Current Assignee: Red Hat, Inc. (International Business Machines Corporation)
• Original Assignee: Red Hat, Inc. (International Business Machines Corporation)
• Inventors: Whitehouse, Steven John
• Primary Examiner(s): VU, BAI DUC
• Application Number:
  US15/206,105
• Publication Number:
  US 20160321292A1
• Time in Patent Office: 557 Days
• CPC Class Codes: G06F 16/1774 Locking methods, e.g. locking methods for file systems allowing shared and concurrent access to files
• CPC Class Codes: G06F 16/182 Distributed file systems
• CPC Class Codes: G06F 9/526 Mutual exclusion algorithms

Abstract

A system and method are disclosed for detecting and handling resource contention in a cluster file system. In one implementation, a processing device determines a measure of congestion for a resource that is shared by a first node of a cluster file system and a second node of the cluster file system, where the first node has a first local queue for lock requests for the resource, and wherein the second node has a second local queue for lock requests for the resource. The processing device adjusts a parameter for a node of the cluster file system in view of the measure of congestion.

Claims

1. A method comprising:

tracking, by a processing device, lock requests for a resource shared by a plurality of nodes of a cluster file system, wherein each node of the cluster file system has a local queue, and wherein the respective local queue of each node tracks lock requests for the resource;

maintaining, by the processing device, a plurality of request queue lengths for the local queues of the nodes, wherein a queue length of the plurality of request queue lengths for the local queue associated with the corresponding node comprises a number of lock requests queued for the local node;

determining, by the processing device, a measure of congestion for the resource, wherein the measure of congestion is in view of an average queue length of the plurality of request queue lengths;

determining, by the processing device, a relative congestion factor for a first node of the plurality of nodes; and

adjusting, by the processing device, a relative lock hold time for the first node of the cluster file system in view of the measure of congestion and the relative congestion factor for the first node.

2. The method of claim 1 further comprising:

granting a lock on the resource to the first node;

incrementing, prior to releasing the lock, a first global counter associated with the resource; and

updating, upon release of the lock, a second global counter associated with the resource.

3. The method of claim 2 further comprising:

granting, to a task executing on the first node, while the lock is held by the first node, read access to at least one of the first global counter or the second global counter.

4. The method of claim 2 wherein the updating increments the second global counter by the queue length for the local queue associated with the first node.

5. The method of claim 2 wherein the adjusting of the relative lock hold time for the first node decreases the relative lock hold time in response to an increase in the measure of congestion.

6. An apparatus comprising:

a memory to store a lock for a resource that is shared by a first node of a cluster file system and a second node of the cluster file system, wherein the first node has a first local queue for lock requests for the resource, and wherein the second node has a second local queue for lock requests for the resource; and

a processing device, operatively coupled to the network interface, to;

maintain a first local queue length of the first local queue and a second local queue length of the second local queue, wherein the first queue length for the first local queue is a number of lock requests queued for the first local node, and the second queue length for the second local queue is a number of lock requests queued for the second local node;

determine a measure of congestion for the resource, wherein the measure of congestion is determined in view of an average of the first local queue length and second local queue length;

determine a relative congestion factor for the first node; and

adjust a parameter for the first node of the cluster file system in view of the measure of congestion and the relative congestion factor for the first node.

7. The apparatus of claim 6 wherein the processing device is further to:

grant a lock on the resource to the first node;

increment, prior to releasing the lock, a first global counter associated with the resource; and

update, upon release of the lock, a second global counter associated with the resource.

8. The apparatus of claim 7 wherein the processing device is further to:

grant to a task executing on the first node, while the lock is held by the first node, read access to at least one of the first global counter or the second global counter.

9. The apparatus of claim 7 wherein to update the second global counter, the processing device is to increment the second global counter by the first local queue length.

10. The apparatus of claim 6 wherein the parameter is a relative lock hold time.

11. The apparatus of claim 10 wherein the adjusting of the relative lock hold time for the first node increases the relative lock hold time in response to a decrease in the measure of congestion.

12. The apparatus of claim 10 wherein the adjusting of the relative lock hold time for the first node decreases the relative lock hold time in response to an increase in the measure of congestion.

13. The apparatus of claim 6 wherein to determine the relative congestion factor for the first node, the processing device is further to:

determine a current length of the first local queue;

determine a current length of the second local queue; and

divide the current length of the first local queue by the average of the first local queue length and second local queue length.

14. A non-transitory computer-readable storage medium having instructions stored thereon that, when executed by a processing device, cause the processing device to:

maintain a first local queue length of a first local queue and a second local queue length of a second local queue, wherein the first queue length for the first local queue is a number of lock requests queued for the first local node, and the second queue length for the second local queue is a number of lock requests queued for the second local node;

determine, by the processing device, a measure of congestion for a resource that is shared by a first node of a cluster file system and a second node of the cluster file system, wherein the first node has the first local queue for lock requests for the resource, wherein the second node has the second local queue for lock requests for the resource, and wherein the measure of congestion is in view of an average of the first local queue length and second local queue length;

determine a relative congestion factor for the first node; and

adjust a parameter for the first node of the cluster file system in view of the measure of congestion and the relative congestion factor for the first node.

15. The non-transitory computer-readable storage medium of claim 14, wherein the instructions further cause the processing device to:

grant a lock on the resource to the first node;

increment, prior to releasing the lock, a first global counter associated with the resource; and

update, upon release of the lock, a second global counter associated with the resource.

16. The non-transitory computer-readable storage medium of claim 15, wherein the instructions further cause the processing device to:

grant to a task executing on the first node, while the lock is held by the first node, read access to at least one of the first global counter or the second global counter.

17. The non-transitory computer-readable storage medium of claim 15, wherein to update the second global counter, the processing device is to increment the second global counter by the first local queue length.

18. The non-transitory computer-readable storage medium of claim 14, wherein the parameter is a relative lock hold time.