[Perforce] Admin Workshop
1 like416 views
This document discusses virtualization, monitoring, and replication of Perforce software. It provides recommendations for virtualizing Perforce, such as using vSphere 5 with certain network drivers. It also discusses monitoring Perforce processes and logs to detect performance issues. New replication features in upcoming Perforce releases are outlined, including commit/edge servers that allow work to be distributed across multiple VMs.
![9
Monitoring
#!/bin/bash
count=$(ps -ef | grep -i p4d_1 | wc –l)
Load=100 # Set to 3 - 4 times your average load.
if [ "$count" -gt "$load" ]
then
echo "Load alert is set to: ${load}." > count.txt
mail -s "p4 count is ${count}.” you@company.com < count.txt
rm count.txt
fi](https://image.slidesharecdn.com/admin-workshop-slides-130523174141-phpapp02/85/Perforce-Admin-Workshop-9-320.jpg)
[Perforce] Admin Workshop
- 1. 1 Virtualization, Monitoring & Replication Workshop Rusty Jackson Sr. Consultant Perforce Software
- 3. 3 Virtualization • Performance should be acceptable if: • Well managed / well planned • Hardware recommendations are the same as bare metal to get the best performance if the installation demands it • Fast CPU preferred over more cores • Memory to fully cache db • Fast storage for the db files
- 4. 4 Virtualization • Use vSphere 5 with vmxnet3 1.0.36 or later for best performance • Performance varies with hardware and VM configuration just like it does on bare metal • http://www.vmware.com/files/pdf/vmw-Perforce- Server-on-VMware-vSphere.pdf
- 5. 5 Virtualization • Perforce is evolving with the market • Perforce works well in a virtualized environment right now for the majority of our customers • The Commit/Edge server functionality coming in the 2013.2 release will allow even our largest customers to make better use of VM’s by distributing the workload across multiple VM’s • 2013.3 – lockless reads on the database
- 6. 6 Monitoring
- 7. 7 Monitoring • Turn monitoring on: • p4 configure set monitor=1 (or 2) • 1 = monitor active processes • 2 = monitor both active and idle processes • p4 monitor show –a –e • -a = show command args • -e = show command environment
- 8. 8 Monitoring • Create a cron job to monitor your p4d process count • You will have to adjust the alert level as you learn what the average load for your server is
- 9. 9 Monitoring #!/bin/bash count=$(ps -ef | grep -i p4d_1 | wc –l) Load=100 # Set to 3 - 4 times your average load. if [ "$count" -gt "$load" ] then echo "Load alert is set to: ${load}." > count.txt mail -s "p4 count is ${count}.” you@company.com < count.txt rm count.txt fi
- 10. 10 Monitoring • Log analysis • Support can assist with a log analysis if you are having performance issues • Log file needs to include debug info. • p4 configure set server=3 • Analyzing your own logs • https://kb.perforce.com/psla
- 11. 11 Monitoring • Structured logs • Command delimited format, easier to parse • p4 configure set serverlog.file.n=logname.csv • p4 configure set serverlog.retain.n=# • 1 = all • 2 = commands • 3 = errors • 4 = audit • 5 = track • 6 = user • 7 = events
- 12. 12 Replication
- 13. 13 Replication • Types of replicas today • Read only • Forwarding • Build Farm • 2013.1 New replication features • On demand library setting • Replica of a replica
- 14. 14 Replication • 2013.2 - Commit and Edge Servers • Commit server is a central repository for files • Edge servers do the vast majority of the work • Workspaces move to the edge servers • Supports horizontal growth of your Perforce installation • Checkpoints and backup required for both the commit and the edge servers
- 15. 15 Replication • High level steps to create a replica • Create configure entries, service user and group and update protect table on master • Checkpoint the master • Copy the checkpoint and version files to the replica • Restore the checkpoint on the replica • Start the server on the replica using the settings from the configure table
- 16. 16 On to the demo… consulting@perforce.com