![]() ![]() These queries can then either be searched on the SAP search or you can check to see if these are your own custom queries which need to be looked at it terms of optimization. The dumps will show you what exact queries were running at the time of the hang / freeze. See Troubleshooting High CPU for further info.Īs the HANA Studio will more than likely be hanging during the hang / slow performance, you can use SSH to run the Runtime Dumps with 2 minute intervals through the means of this hdbsql script:ĭATE=`date +%Y%m%d.%H%M%S` for PID in `ps x|grep -E “hdb(index)server”|grep -v grep|awk ‘.dump doneĪfter running the script you can then open the generated RTE dumps. These threads then link you back to short stacks which then can checked to see what exactly this stack is doing. The thread of this SQL query can then be checked via the parent and also its child threads. Large Query plans can indicate problematic SQL Queries. Runtime Dumps can useful for the following situations:īy checking a RTE dump you can look for certain key indicators such as Large Query plans. The first thing you have to think of when you face this situation is to execute Runtime Dumps immediately. THP will make HANA run quicker for a while but when it comes to splitting THP’s, the system gets so slow that working with it is not possible. Transparent Huge Pages must be deactivated. The most probable reason for this is outside HANA. If this is high, then we have to find out why?. ![]() The remaining cache size is (in general) not available for HANA. When checking the cache size (top command) please remember that HANA shared memory is also booked as “shared”. ![]() So it is important that HANA is really assigned the memory up to the GAL and nothing else using this.Ī large file cache can lead to problem. Other non HANA processes or other instances can interfere with the HANA system. What you have to remember is that HANA limits its memory usage by the Global Allocation Limit. Check the Swap, File Cache and the Transparent Hugepages.(Either wrong OS configuration or an OS/Hardware issue.) Traces needed for further analysis: Runtime Dumps.ģ: Hanging situations where a logon via SSH is not possible.Traces needed for further analysis : Runtime Dumps + Kernel Profiler.Ģ: Hanging situations with CPU Utilization because all threads are waiting.So here are some issues that you may face with hanging and what Traces are needed to troubleshoot.ġ: Hanging situation with High CPU Utilization. Below is a small example of troubleshooting this issue. So a small example of a hang / freeze situation I have witnessed is when a user goes to open the “Administrative Tab” in HANA Studio and the system hangs for a long period of time. Even when looking at the most basic single sever, single host, we can see many areas which over lap and interact with one another, in other words, when you see a hang do not just assume the cause is all related to HANA. Like all other software systems, SAP HANA relies on hardware to run its processes. Application is hanging but new connections to the database are still possible (Possibly not a HANA DB issue which must be analysed from an application perspective before looked at from a HANA DB perspective). ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |