Every now and then we, as a PureCloud partner, see customers run into server management snags that could be prevented. I’ve picked a handful of common scenarios to examine.
Rule number 1. Manage manual maintenance switchovers monthly. Mmmmkay?
PureConnect Switchover is a robust Active-Active redundant server strategy that over the years has slowly evolved to include more and more subsystems that it can keep in sync between the two PureConnect servers. What has not changed is the need to make sure you have a fresh backup server ready and waiting in the wings in the event that something on the Primary PureConnect server goes wrong. Genesys and CPI.Solutions recommend that you keep a backup server refreshed every 30 days by doing a manual switchover. Doing this ensures several things:
- The backup server is actually replicating
- The backup server is actually working
- You ensure that nothing has changed under your virtual guest machine that could invalidate your license
What does this look like?
Let’s say your system has been up and running for the last 29 days. I would suggest the following steps:
- Reboot the backup PureConnect server.
- Verify that the server restarts and replicates with the Primary PureConnect by checking in the Switchover control panel. If it doesn’t show up as the backup server, something has gone wrong and will need to be investigated.
- Manually switchover, preferably during off-peak hours. While connected calls will not be disconnected and will continue as though nothing happened, there is generally a period of between 30-60 seconds that new calls placed in and out of the system will receive a busy signal.
- Give the newly active server a few moments to fully shut down its connections to the now stale PureConnect server.
- Reboot the stale PureConnect server.
- Once again, re-verify that the server restarts and replicates with the Switchover Control panel. If it fails to properly connect after 20 minutes, it should be properly investigated.
- Congratulations! You can take confidence that your PureConnect servers are handling switchover correctly. Resume reading your copy of Crime and Punishment in the original pre-revolutionary Russian.
Free the Memory!
When your PureConnect Server was young and new it may not have had many chores. But over time, a few more departments have been added and the system is doing a lot more work. Genesys starts with a recommended 4 GB of RAM for each PureConnect server. It doesn’t take much additional loading before the server could start showing higher RAM usage. We recommend that you occasionally take a look at your free system RAM percentage. While high usage spikes might occur, keep the average free memory percentage at 50%. Doing this will help prevent degraded performance under heavy loading.
Keep those disks clear!
PureConnect in a standard configuration will use at least 3 different logical hard drive volumes. A drive dedicated to the operating system, a drive dedicated to the PureConnect application and a third dedicated to application logging.
It goes without saying (but I’m saying it anyway), bad things can happen if your drives fill up. Yes, if you let the OS drive fill up, or the volume where your pagefile resides, your system will grind to a halt. Don’t let that happen!
Likewise, if you let the PureConnect application volume fill up, the PureConnect application will halt. Don’t let that happen! While this is not common in the modern world of cheap-ish disk storage, it is still possible. What sometimes occurs is that PMQ files or temporary recorder files don’t get processed and fill up remaining drive space.
The 3rd scenario is the logging volume fills up. When this happens, the system continues to run, but logging does not occur. Without logging, we are usually unable to determine the cause of other more severe systemic issues. You should have enough free logging space to keep 8 days of logs at standard interaction volume at standard trace levels.
In addition, you should keep additional space free to sustain enhanced trace levels or higher interaction volumes for two to three days at a minimum or longer. Log files are zip compressed every night, so a day’s worth of uncompressed logs will usually shrink by a factor of 10 or more. The big five logs (IP, Notifier, TelephonyServices, SIPEngine and Client Services) can easily take 10s of gigabytes each if their trace levels need to be elevated. As a rule of thumb, I would consider keeping 50 to 100% additional free space on hand in the event it needs to be called upon. A heavily utilized system could require higher reserves.
And a final word about system backups. Backup.
Next time, we will discuss what to do with PMQ .err files, orphaned recordings and some log file management tips.
Thanks for reading!