When upgrading large applications, your system can be at risk of overloading. Such upgrades can produce a big traffic for the OSS database.
Hence, if you plan to upgrade large-scale applications (over 10,000 subscriptions), a proper maintenance window should be considered.
Adhere the following guidelines to prepare for such upgrades:
- Minimize amount of UI sessions.
- Make sure that there are no
Synchronize resource usage for ...
periodic task running or such tasks will not be started in the nearest two hours. -
Daily Billing Process
is not running and will not be started in the nearest two hours. - There are no APS API / XMLRPC calls to OSS.
- OSS and OSS database CPU and disk load is minimal.
The following examples would be helpful to receive information on the aforementioned items.
To get amount of UI sessions for the period of one hour:
# sed -n '/^00:/,/^01:/p' /var/log/pa/pui/pui.log | grep 'cleaning task' ... Session cleanup: result of static static storages cleaning task: alive=1 hibernated=0 killed=0 zombie=0 ...
To get the number of XMLRPC calls to OSS for a particular period:
# kubectl logs oss-node-0 |sed -n '/^Aug 13 01/,/^Aug 13 02/p' | grep -F '[openapi:' | grep 'checking that remote host' | wc -l
To get the number of APS API calls to OSS for a particular period:
# kubectl logs oss-node-0 |sed -n '/^Aug 13 01/,/^Aug 13 02/p' | grep -F '[c.p.p.s.a.RootRoutingServlet request from:' | grep 'checking that remote host' | wc -l