Verify Network Communications for Cluster Discovery
If you want to use the discover cluster capabilities in Parallel Computing Toolbox™, your network must be configured to use DNS SRV or DNS TXT records.
DNS SRV Record
When you use DNS for MATLAB® Job Scheduler cluster discovery, you require a DNS SRV record for each domain. You can have multiple DNS SRV records for multiple MATLAB Job Schedulers. Use the following general form for each DNS SRV record.
_mdcs._tcp.<domain> <TTL> IN SRV <priority> <weight> <port> <hostname>.Construct a DNS SRV record for a MATLAB Job Scheduler server using the following parts.
<domain>is the domain name (likecompany.comoruniversity.edu) that the client machine searches.<TTL>indicates how long (in seconds) the DNS record can be cached.3600is recommended.IN SRVis required as shown, indicating that this is a service record.<priority>and<weight>indicate priority and weight values. If you create multiple DNS SRV records, you can specify their priority with these fields. A value of0is recommended for each. The lower<priority>is, the higher priority the host has. When two records have the same<priority>, the record with the highest<weight>is used first. Use the<weight>value to specify server preference.<port>is the port on which you connect to the MATLAB Job Scheduler server. The default port is27350. If you change port for the MATLAB Job Scheduler server, change<port>accordingly.<hostname>is the fully qualified domain name for the host serving the MATLAB Job Scheduler. The machinemjs-1on the domaincompany.comhas a fully qualified domain namemjs-1.company.com.
A valid DNS SRV record for the company.com network running a
MATLAB Job Scheduler on machine mjs-1 might look like this:
_mdcs._tcp.company.com 3600 IN SRV 0 0 27350 mjs-1.company.com.
Note
If multiple domains are required to locate the cluster, use a DNS SRV record for each domain. If the network accessed by users via VPN has different DNS SRV records to your internal network, ensure that a DNS SRV record exists for each domain.
Use the standard procedure for your DNS system to create appropriate DNS SRV records.
You can use standard utilities such as the nslookup command to verify
that your network is configured with the necessary DNS SRV records. To examine MATLAB Job Scheduler DNS SRV records for the company.com domain,
use the following command.
nslookup -type=SRV _mdcs._tcp.company.com DNS TXT Record
Use DNS TXT records for third-party scheduler cluster discovery. A DNS TXT record associates a text string with a particular domain. To let MATLAB know where to find cluster discovery configuration files, store the locations of cluster discovery configuration files as text strings in DNS TXT records.
You can have multiple DNS TXT records for multiple clusters. Use this general form for each DNS TXT record.
_mdcs._tcp.<domain> IN TXT "discover_folder=<folder>"Construct a DNS TXT record to discover a third-party scheduler using these parts.
<domain>is the domain name (likecompany.comoruniversity.edu) that the client machine searches.IN TXTis required as shown, indicating that this is a text record."discover_folder=<folder>"where <folder> is the location of your cluster discovery configuration files.
A valid DNS TXT record for the company.com network running a Slurm
scheduler cluster with a cluster discovery configuration file stored in
/network/share/discovery might look like this:
_mdcs._tcp.company.com IN TXT "discover_folder=/network/share/discovery"Note
If multiple domains are required to locate the cluster, use a DNS TXT record for each domain. If the network accessed by users via VPN has different DNS TXT records to your internal network, ensure that a DNS TXT record exists for each domain.
Use the standard procedure for your DNS system to create appropriate DNS TXT records.
You can use standard utilities such as the nslookup command to verify
that your network is configured with the necessary DNS TXT records. To examine DNS TXT
records for the company.com domain, use the following command.
nslookup -type=TXT _mdcs._tcp.company.comSee Also
Topics
- Discover Clusters and Use Cluster Profiles (Parallel Computing Toolbox)
- Configure for Third-Party Scheduler Cluster Discovery