CN110177020A - A kind of High-Performance Computing Cluster management method based on Slurm - Google Patents
A kind of High-Performance Computing Cluster management method based on Slurm Download PDFInfo
- Publication number
- CN110177020A CN110177020A CN201910524257.XA CN201910524257A CN110177020A CN 110177020 A CN110177020 A CN 110177020A CN 201910524257 A CN201910524257 A CN 201910524257A CN 110177020 A CN110177020 A CN 110177020A
- Authority
- CN
- China
- Prior art keywords
- node
- slurm
- cluster
- calculate
- management method
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/06—Management of faults, events, alarms or notifications
- H04L41/0654—Management of faults, events, alarms or notifications using network fault recovery
- H04L41/0663—Performing the actions predefined by failover planning, e.g. switching to standby network elements
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/08—Configuration management of networks or network elements
- H04L41/0893—Assignment of logical groups to network elements
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/50—Network service management, e.g. ensuring proper service fulfilment according to agreements
- H04L41/5041—Network service management, e.g. ensuring proper service fulfilment according to agreements characterised by the time relationship between creation and deployment of a service
- H04L41/5054—Automatic deployment of services triggered by the service manager, e.g. service implementation by automatic configuration of network components
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L43/00—Arrangements for monitoring or testing data switching networks
- H04L43/10—Active monitoring, e.g. heartbeat, ping or trace-route
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/2866—Architectures; Arrangements
- H04L67/30—Profiles
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Health & Medical Sciences (AREA)
- Cardiology (AREA)
- General Health & Medical Sciences (AREA)
- Hardware Redundancy (AREA)
Abstract
The High-Performance Computing Cluster management method based on Slurm that the present invention relates to a kind of, comprising: an arbitrarily selected machine is as control node, and other machines are as calculate node；The host name of all calculate nodes or IP information in cluster are got, copies cluster installation kit and installation script to each calculate node；Each calculate node is arrived by SSH service login in control node, deployment is built by cluster environment on installation script completion node；Process is controlled and received in control node deployment, monitors computational resource is used for, receives the information that calculate node is sent；There are finger daemons to be used to manage calculate node in cluster in each calculate node, information and is sent to control node by SSH service in timing acquiring node state and node；Carry out coordinated management to calculate node and redundancy backup node；Based on Slurm job management mechanism and node state monitoring process, the present situation according to cluster system resource disposes the operation in queue, monitors and batch operation.
Description
Technical field
The present invention relates to data analysis management technology, a kind of management method of the High-Performance Computing Cluster based on Slurm framework.
Background technique
In current informationized society, the analysis of mass data and processing problem propose very greatly the computing capability of system
Challenge.But the limitation of existing computing platform makes calculating service be unable to Dynamic Integration, and data storage, data is caused to calculate analysis
Etc. there are certain bottlenecks.Multiple computing resources are constituted group system by network interconnection to have been found to be to solve the problems, such as this
Effective means.However cluster scale becomes larger, structure is complicated, calculates high isomerism and the application pair of environment for caused cluster topology
The diversification of the nonfunction requirements such as performance greatly increases the processes difficulty such as installation, configuration and management of group system.Cluster
System is reduced its management cost by Self management mode, adapted to there is an urgent need to change current static state, passively way to manage
Calculate the personalized nonfunction requirement of variation and the application of environment.
Summary of the invention
The High-Performance Computing Cluster management method based on Slurm that the purpose of the present invention is to provide a kind of, it is above-mentioned existing for solving
There is the problem of technology,
A kind of High-Performance Computing Cluster management method based on Slurm of the present invention, comprising: an arbitrarily selected machine is as control
Node, other machines are as calculate node；The host name of all calculate nodes or IP information in cluster are got, cluster is installed
Packet and installation script copy each calculate node to；Passed through by SSH service login to each calculate node in control node
Cluster environment builds deployment on installation script completion node；Process is controlled and received in control node deployment, is calculated for monitoring
Resource receives the information that calculate node is sent；It is used to manage the calculating in cluster there are finger daemon in each calculate node
Node information and is sent to control node by SSH service in timing acquiring node state and node；To calculate node and superfluous
The carry out coordinated management of remaining backup node；Based on Slurm job management mechanism and node state monitoring process, according to group system
The present situation of resource disposes the operation in queue, monitors and batch operation.
One embodiment of the High-Performance Computing Cluster management method according to the present invention based on Slurm, wherein node fault-tolerance backup
Method includes: the deployment Slurmctld finger daemon in control node, the configuration file of timing acquisition Slurm and is saved before
Status information, disk is periodically written into the status information of full controller, if control node break down, incremental changes letter
Disk can be written in breath immediately, and system completes control according to the Standby control node being arranged in advance in configuration file slurm.conf at this time
The switching of node processed；Dispose Slurmd finger daemon in calculate node, monitor physical computing nodes and operating system, Slurmd into
Journey is based on heartbeat inspecting mechanism feedback and reports calculate node state.
One embodiment of the High-Performance Computing Cluster management method according to the present invention based on Slurm, wherein in control node,
Starting one monitoring finger daemon of maintenance, and operation is interacted, starts, hangs up and cancelled by api interface and Slurm；
In calculate node, according to the heartbeat message agreement agreed upon, periodically to the health monitoring finger daemon for being located at control node
Heartbeat message is sent, error message is reported.
One embodiment of the High-Performance Computing Cluster management method according to the present invention based on Slurm, wherein if application program goes out
Wrong perhaps operation time-out is by restarting this operation or by job migration to spare calculate node；If physical node delay machine,
Close this node, and by job migrations whole on this node to backing up calculate node.
One embodiment of the High-Performance Computing Cluster management method according to the present invention based on Slurm, wherein from the configuration of cluster
The host name of all calculate nodes or IP information in cluster are got in file slurm.conf.
One embodiment of the High-Performance Computing Cluster management method according to the present invention based on Slurm, wherein instructed by scp
Copy cluster installation kit and installation script to each calculate node.
One embodiment of the High-Performance Computing Cluster management method according to the present invention based on Slurm, wherein setting clustered control
Password login is exempted from by SSH between node and each calculate node.
One embodiment of the High-Performance Computing Cluster management method according to the present invention based on Slurm, wherein on all the nodes
Installation production domesticization milky way kylin system.
One embodiment of the High-Performance Computing Cluster management method according to the present invention based on Slurm, wherein made based on Slurm
Industry administrative mechanism and node state monitoring process, the present situation according to cluster system resource is to the operation carry out portion in queue
Administration, monitoring and batch operation include: that operation deployment includes: by the node state monitoring process in control node
NodeMonitor determines the available computational resources number in cluster, submits the information that will run operation later；Operation control packet
It includes: Operation control being hung up current work when the operation of high priority enters if it exists by scancel order in Slurm
Until high priority job run is completed.As a result extracting includes: that job execution is judged after the completion, if running succeeded, is shown
Show last output as a result, returning to failure cause if failure.
The present invention proposes that a kind of script mounting means completes the deployment for the isomeric group that different architectural nodes are built, significantly simple
The complex operations of isomeric group installation configuration are changed.On this basis, the present invention provides a kind of management method of isomeric group, mentions
For modules such as resource management monitoring, fault-tolerant support, job managements, by distributing rationally, Rational provides surely for user jointly
Fixed efficient Heterogeneous Cluster Environment.
Detailed description of the invention
Fig. 1 show isomeric group basic block diagram；
Fig. 2 show key deployment cluster script process figure；
Fig. 3 show resource management monitoring module and realizes interaction figure；
Fig. 4 show node fault-tolerance administrative mechanism schematic diagram；
Fig. 5 show operation deployment flow chart；
Fig. 6 show Operation control process figure；
Fig. 7 show result and extracts flow chart.
Specific embodiment
To keep the purpose of the present invention, content and advantage clearer, with reference to the accompanying drawings and examples, to of the invention
Specific embodiment is described in further detail.
A kind of High-Performance Computing Cluster management method based on Slurm, comprising:
1. selecting multiple computer nodes builds High Performance Cluster System, the hardware structure of node is unlimited.In all nodes
Upper installation production domesticization milky way kylin system, Fig. 1 show isomeric group basic block diagram, as shown in Figure 1.
2. the deployment of building of High Performance Cluster System is a complicated process, a kind of script peace is provided based on Slurm
Dress mode, Slurm are a kind of telescopic in height that can be used for mass computing node cluster and fault-tolerant cluster manager dual system and operation
Scheduling system, Fig. 2 show key deployment cluster script process figure, as shown in Figure 2.
(1) for an arbitrarily selected machine as control node, clustered control section is arranged as calculate node in other machines
It can be in order to avoid password login by SSH between point and each calculate node；
(2) host name of all calculate nodes or IP letter in cluster are got from the configuration file slurm.conf of cluster
Breath is instructed by scp and copies cluster installation kit and installation script to each calculate node；
(3) it by SSH service login to each calculate node in control node, completes to collect on node by installation script
Build deployment in group rings border；
By this script mounting means, largely reduce the troublesome operation of cluster installation, user can pass through control
Node processed completes the configuration installation of all nodes in cluster.
3. High-Performance Computing Cluster management method proposed by the present invention mainly passes through resource management monitoring module, fault-tolerant support mould
Block, job management module etc. are realized.
(1) resource management monitoring module
Fig. 3 show resource management monitoring module and realizes interaction figure, as shown in figure 3, merging in resource management monitoring module
Entire group system and Internet resources construct shared, configurable resource pool, provide all computing resources in system
Flexible monitoring management, High Availabitity guarantee and automation O&M.
The critical component status information that each node in cluster can be got in the present invention in real time, in control node
Control receiving process is disposed, monitors computational resource is used for, receives the information that calculate node is sent；In each calculate node
There are finger daemon monitor, monitor is used to manage the calculate node in cluster, its timing acquiring node state and section
The relevant information of the critical components such as CPU, memory, network, disk and control node is sent to by SSH service on point.
(2) fault-tolerant support module
Fig. 4 show node fault-tolerance administrative mechanism schematic diagram, as shown in figure 4, fault-tolerant support module is mainly used for calculating
The coordinated management of node and redundancy backup node, solve due to individually calculate or control unit fail caused by the system failure.
Slurm node administration mechanism is merged in the present invention and application program intelligent management mechanism proposes that a kind of new node holds
Wrong backup method.
Step 1: in control node dispose Slurmctld finger daemon, the configuration file of timing acquisition Slurm and it
Disk is periodically written in the status information of full controller by the status information of preceding preservation, if control node breaks down, increment
Disk can be written in change information immediately, and system is according to the Standby control node being arranged in advance in configuration file slurm.conf at this time
Complete the automatic switchover of control node.
Step 2: Slurmd finger daemon is disposed in calculate node, monitors physical computing nodes and operating system, Slurmd
Process is based on heartbeat inspecting mechanism feedback and reports calculate node state, reports the states such as physics delay machine or operating system collapse.
Step 3: if Slurm software itself break down or upper layer parallel computation application program obstruction, at this time failure without
Method is monitored.Therefore, an application program intelligent management platform is constructed for monitoring application program.In control node, starting
It safeguards a monitoring finger daemon, and is interacted by api interface and Slurm, start, hang up, cancel operation；It is saved calculating
Point, according to the heartbeat message agreement agreed upon, from intelligent management application program periodically to the health prison for being located at control node
It controls finger daemon and sends heartbeat message, report the error messages such as application program error or operation time-out.
For the failure of appearance, also use two kinds of different faults countermeasures in the present invention, if application program error or
Person's operation time-out, this operation is restarted by application program intelligent management platform or is saved job migration to spare calculating
Point；If physical node delay machine, this node is closed by management platform, and job migrations whole on this node to backup are calculated into section
Point.
(3) job management module
Fig. 5 show operation deployment flow chart, as shown in figure 5, job management module can working as according to cluster system resource
Preceding situation such as is disposed to the operation in queue, is monitored, being distributed at the operation.It is based on Slurm job management mechanism and node shape
State monitoring process etc. is realized, provides the functions such as operation deployment, Operation control, result extraction for user.
1) operation is disposed
Fig. 6 show Operation control process figure, as shown in fig. 6, the main submission for realizing operation of operation deployment, modification and
The operation such as deletion, is determined in cluster by the node state monitoring process NodeMonitor in control node use tricks first
Number of resources is calculated, the relevant information that will run operation is submitted later, mainly there is job title, required computing resource quantity, operation
Number etc..
2) Operation control
Operation control is mainly used for realizing the controls such as hang-up, the release of operation operation.It is ordered by scancel in Slurm etc.
The flexible control realized to operation is enabled, when the operation of high priority enters if it exists, system can be hung up current work until high
Priority job operation is completed.Operation control process is as follows:
3) result is extracted
Fig. 7 show result and extracts flow chart, as shown in fig. 7, it is mainly that user exports final calculating knot that result, which is extracted,
Fruit.Judged after the completion of job execution, if running succeeded, show last output as a result, returning to failure cause if failure,
User is allowed more intuitively to obtain output result.
The present invention proposes a kind of based on Slurm aiming at the problem that management difficult to arrange of current isomery High-Performance Computing Cluster
The management method of the High-Performance Computing Cluster of (Simple Linux Utility for Resource Management) building.It is first
First, the hardware structure of each node of building High-Performance Computing Cluster is determined；Secondly, completing High-Performance Computing Cluster by script mounting means
Build；Finally, providing the dynamic management that the modules such as resource management monitoring, fault-tolerant support, job management carry out cluster.
The configuration management of current high performance cluster is a cumbersome process, and the present invention proposes that a kind of script mounting means is complete
At the deployment for the isomeric group that different architectural nodes are built, the complex operations of isomeric group installation configuration are enormously simplified.Herein
On the basis of, the present invention provides a kind of management method of isomeric group, provides resource management monitoring, fault-tolerant support, job management etc.
Module, by distributing rationally, Rational provide the Heterogeneous Cluster Environment of stability and high efficiency jointly for user.
The above is only a preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, without departing from the technical principles of the invention, several improvement and deformations can also be made, these improvement and deformations
Also it should be regarded as protection scope of the present invention.
Claims (9)
1. a kind of High-Performance Computing Cluster management method based on Slurm, comprising:
An arbitrarily selected machine is as control node, and other machines are as calculate node；
The host name of all calculate nodes or IP information in cluster are got, cluster installation kit and installation script are copied to each
Calculate node；
By SSH service login to each calculate node in control node, cluster environment on node is completed by installation script
Build deployment；
Process is controlled and received in control node deployment, monitors computational resource is used for, receives the information that calculate node is sent；Each
There are finger daemons to be used to manage the calculate node in cluster, information in timing acquiring node state and node in calculate node
And control node is sent to by SSH service；
Carry out coordinated management to calculate node and redundancy backup node；
Based on Slurm job management mechanism and node state monitoring process, the present situation according to cluster system resource is to queue
In operation disposed, monitored and batch operation.
2. the High-Performance Computing Cluster management method based on Slurm as described in claim 1, which is characterized in that node fault-tolerance backup
Method includes:
Slurmctld finger daemon, the configuration file of timing acquisition Slurm and previously stored state are disposed in control node
Disk is periodically written in the status information of full controller by information, if control node breaks down, incremental changes information can be stood
Disk is written, system completes control node according to the Standby control node being arranged in advance in configuration file slurm.conf at this time
Switching；
Slurmd finger daemon is disposed in calculate node, monitors physical computing nodes and operating system, Slurmd process is based on the heart
It jumps monitoring mechanism feedback and reports calculate node state.
3. the High-Performance Computing Cluster management method based on Slurm as claimed in claim 2, which is characterized in that in control node, open
One monitoring finger daemon of dynamic maintenance, and operation is interacted, starts, hangs up and cancelled by api interface and Slurm；?
Calculate node is periodically sent out to the health monitoring finger daemon for being located at control node according to the heartbeat message agreement agreed upon
Heartbeat message is sent, error message is reported.
4. the High-Performance Computing Cluster management method based on Slurm as claimed in claim 3, which is characterized in that if application program goes out
Wrong perhaps operation time-out is by restarting this operation or by job migration to spare calculate node；If physical node delay machine,
Close this node, and by job migrations whole on this node to backing up calculate node.
5. the High-Performance Computing Cluster management method based on Slurm as described in claim 1, which is characterized in that from the configuration of cluster
The host name of all calculate nodes or IP information in cluster are got in file slurm.conf.
6. the High-Performance Computing Cluster management method based on Slurm as described in claim 1, which is characterized in that being instructed by scp will
Cluster installation kit and installation script copy each calculate node to.
7. the High-Performance Computing Cluster management method based on Slurm as described in claim 1, which is characterized in that setting clustered control
Password login is exempted from by SSH between node and each calculate node.
8. the High-Performance Computing Cluster management method based on Slurm as described in claim 1, which is characterized in that on all the nodes
Installation production domesticization milky way kylin system.
9. the High-Performance Computing Cluster management method based on Slurm as described in claim 1, which is characterized in that be based on Slurm operation
Administrative mechanism and node state monitoring process, according to cluster system resource the present situation the operation in queue is disposed,
Monitoring and batch operation include:
Operation deployment includes: to determine can be used in cluster by the node state monitoring process NodeMonitor in control node
Computing resource number submits the information that will run operation later；
Operation control includes: by scancel order in Slurm to Operation control, when the operation of high priority enters if it exists,
Current work is hung up until high priority job run is completed.
As a result extracting includes: that job execution is judged after the completion, if running succeeded, shows last output as a result, if failure
Then return to failure cause.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910524257.XA CN110177020A (en) | 2019-06-18 | 2019-06-18 | A kind of High-Performance Computing Cluster management method based on Slurm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910524257.XA CN110177020A (en) | 2019-06-18 | 2019-06-18 | A kind of High-Performance Computing Cluster management method based on Slurm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110177020A true CN110177020A (en) | 2019-08-27 |
Family
ID=67697432
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910524257.XA Pending CN110177020A (en) | 2019-06-18 | 2019-06-18 | A kind of High-Performance Computing Cluster management method based on Slurm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110177020A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111541591A (en) * | 2020-07-09 | 2020-08-14 | 武汉绿色网络信息服务有限责任公司 | SSH-based server detection method and device |
| CN111949389A (en) * | 2020-08-11 | 2020-11-17 | 曙光信息产业(北京)有限公司 | Slurm-based information acquisition method and device, server and computer-readable storage medium |
| CN112380086A (en) * | 2019-09-29 | 2021-02-19 | 北京城建设计发展集团股份有限公司 | Intelligent sensing control system and method for distributed micro-service architecture data center |
| CN112445595A (en) * | 2020-11-26 | 2021-03-05 | 深圳晶泰科技有限公司 | Multitask submission system based on slurm computing platform |
| CN112737934A (en) * | 2020-12-28 | 2021-04-30 | 常州森普信息科技有限公司 | Cluster type Internet of things edge gateway device and method |
| WO2022109932A1 (en) * | 2020-11-26 | 2022-06-02 | 深圳晶泰科技有限公司 | Multi-task submission system based on slurm computing platform |
| CN114584455A (en) * | 2022-03-04 | 2022-06-03 | 吉林大学 | Small and medium-sized high-performance cluster monitoring system based on enterprise WeChat |
| CN114745385A (en) * | 2022-04-12 | 2022-07-12 | 吉林大学 | Method for constructing slurm scheduling parallel computing cluster |
| CN115202992A (en) * | 2022-09-15 | 2022-10-18 | 中国空气动力研究与发展中心计算空气动力研究所 | CFD operation convergence monitoring method for slurm scheduling system |
| CN115834594A (en) * | 2022-11-16 | 2023-03-21 | 贵州电网有限责任公司 | Data collection method for improving high-performance computing application |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103167041A (en) * | 2013-03-28 | 2013-06-19 | 广州中国科学院软件应用技术研究所 | System and method for supporting cloud environment application cluster automation deployment |
| CN105100267A (en) * | 2015-08-24 | 2015-11-25 | 用友网络科技股份有限公司 | Deployment apparatus and deployment method for large enterprise private cloud |
| CN106844021A (en) * | 2016-12-06 | 2017-06-13 | 中国电子科技集团公司第三十二研究所 | Computing environment resource management system and management method thereof |
| CN108319514A (en) * | 2018-01-26 | 2018-07-24 | 山东超越数控电子股份有限公司 | A kind of visual scheduling system based on Slurm job managements |
-
2019
- 2019-06-18 CN CN201910524257.XA patent/CN110177020A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103167041A (en) * | 2013-03-28 | 2013-06-19 | 广州中国科学院软件应用技术研究所 | System and method for supporting cloud environment application cluster automation deployment |
| CN105100267A (en) * | 2015-08-24 | 2015-11-25 | 用友网络科技股份有限公司 | Deployment apparatus and deployment method for large enterprise private cloud |
| CN106844021A (en) * | 2016-12-06 | 2017-06-13 | 中国电子科技集团公司第三十二研究所 | Computing environment resource management system and management method thereof |
| CN108319514A (en) * | 2018-01-26 | 2018-07-24 | 山东超越数控电子股份有限公司 | A kind of visual scheduling system based on Slurm job managements |
Non-Patent Citations (2)
| Title |
|---|
| 任金龙: "基于Ansible 的云平台自动化部署的研究与实现", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
| 刘杨兵: "基于集群环境的作业管理中间件的研究与实现", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112380086A (en) * | 2019-09-29 | 2021-02-19 | 北京城建设计发展集团股份有限公司 | Intelligent sensing control system and method for distributed micro-service architecture data center |
| CN111541591B (en) * | 2020-07-09 | 2020-09-15 | 武汉绿色网络信息服务有限责任公司 | SSH-based server detection method and device |
| CN111541591A (en) * | 2020-07-09 | 2020-08-14 | 武汉绿色网络信息服务有限责任公司 | SSH-based server detection method and device |
| CN111949389A (en) * | 2020-08-11 | 2020-11-17 | 曙光信息产业(北京)有限公司 | Slurm-based information acquisition method and device, server and computer-readable storage medium |
| CN112445595B (en) * | 2020-11-26 | 2022-10-25 | 深圳晶泰科技有限公司 | Multitask submission system based on slurm computing platform |
| CN112445595A (en) * | 2020-11-26 | 2021-03-05 | 深圳晶泰科技有限公司 | Multitask submission system based on slurm computing platform |
| WO2022109932A1 (en) * | 2020-11-26 | 2022-06-02 | 深圳晶泰科技有限公司 | Multi-task submission system based on slurm computing platform |
| CN112737934A (en) * | 2020-12-28 | 2021-04-30 | 常州森普信息科技有限公司 | Cluster type Internet of things edge gateway device and method |
| CN114584455A (en) * | 2022-03-04 | 2022-06-03 | 吉林大学 | Small and medium-sized high-performance cluster monitoring system based on enterprise WeChat |
| CN114584455B (en) * | 2022-03-04 | 2023-06-30 | 吉林大学 | Small and medium-sized high-performance cluster monitoring system based on enterprise WeChat |
| CN114745385A (en) * | 2022-04-12 | 2022-07-12 | 吉林大学 | Method for constructing slurm scheduling parallel computing cluster |
| CN114745385B (en) * | 2022-04-12 | 2023-05-30 | 吉林大学 | Method for constructing slurm scheduling parallel computing cluster |
| CN115202992B (en) * | 2022-09-15 | 2022-11-22 | 中国空气动力研究与发展中心计算空气动力研究所 | CFD operation convergence monitoring method for slurm scheduling system |
| CN115202992A (en) * | 2022-09-15 | 2022-10-18 | 中国空气动力研究与发展中心计算空气动力研究所 | CFD operation convergence monitoring method for slurm scheduling system |
| CN115834594A (en) * | 2022-11-16 | 2023-03-21 | 贵州电网有限责任公司 | Data collection method for improving high-performance computing application |
| CN115834594B (en) * | 2022-11-16 | 2024-04-19 | 贵州电网有限责任公司 | Data collection method for improving high-performance computing application |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110177020A (en) | A kind of High-Performance Computing Cluster management method based on Slurm | |
| CN107544839B (en) | Virtual machine migration system, method and device | |
| US8364460B2 (en) | Systems and methods for analyzing performance of virtual environments | |
| CN108270726B (en) | Application instance deployment method and device | |
| US20130013766A1 (en) | Computer cluster and method for providing a disaster recovery functionality for a computer cluster | |
| WO2017067484A1 (en) | Virtualization data center scheduling system and method | |
| JP5305040B2 (en) | Server computer switching method, management computer and program | |
| CN108347339B (en) | Service recovery method and device | |
| CN110134518A (en) | A kind of method and system improving big data cluster multinode high application availability | |
| KR20200078328A (en) | Systems and methods of monitoring software application processes | |
| CN110580198B (en) | Method and device for adaptively switching OpenStack computing node into control node | |
| CN110445662A (en) | OpenStack control node is adaptively switched to the method and device of calculate node | |
| CN110958311A (en) | YARN-based shared cluster elastic expansion system and method | |
| CN113626280B (en) | Cluster state control method and device, electronic equipment and readable storage medium | |
| CN112948063A (en) | Cloud platform creation method and device, cloud platform and cloud platform implementation system | |
| CN112199178A (en) | Cloud service dynamic scheduling method and system based on lightweight container | |
| CN112612635B (en) | Multi-level protection method for application program | |
| CN110046064B (en) | Cloud server disaster tolerance implementation method based on fault drift | |
| CN106959885A (en) | A kind of virtual machine High Availabitity realizes system and its implementation | |
| Moghaddam et al. | Self-healing redundancy for openstack applications through fault-tolerant multi-agent task scheduling | |
| US10768996B2 (en) | Anticipating future resource consumption based on user sessions | |
| CN114598591B (en) | Embedded platform node fault recovery system and method | |
| Fu et al. | Proactive resource management for failure resilient high performance computing clusters | |
| Kitamura et al. | Development of a Server Management System Incorporating a Peer-to-Peer Method for Constructing a High-availability Server System | |
| Azaiez et al. | Hybrid fault tolerance model for cloud dependability |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190827 |