Hello everyone.
I've tried to change those parameters and also give more memory to the JVM by modifying the JAVA_OPTS variable in the sampleStartup.sh file. The performance has improved a little bit but I'm not getting the desired one.
Things get worse when I compare it to the old Archive Engine, which running on another VM of similar specifications, has better performance.
In particular, I'm running both services in two different VM but with similar specifications.
- SO: Rocky Linux 9.5.
- Number of CPUs: 4.
- RAM: 8 GB.
- Primary disk with 40 GB.
- Secondary disk with 256 GB in which "/var" is mounted.
With that, I'm able to archive 5000 sinusoidal PVs updating every 500 ms in the VM with the Archive Engine but I'm experiencing data losses with the Archiver Appliance. Also, the VM with the Archiver Appliance is getting full of memory in the root partition even though I have defined the short, medium and long term folders on the "/var" directory, which is mounted on the secondary disk.
I'm thinking that maybe the comparison is unfair with the Archiver Appliance, because the documentation describes that it's possible to deploy multiple appliances, but how can this be done? On each device in the cluster a full instance of the Archiver Appliance is deployed? Or each of the WAR Tomcats can be distributed into different devices? Because inside the appliance.xml there are fields for the IP address of each system (ETL, management, engine & retrieval).
And finally, is there a guide that describes how to perform this complex deployment?
Thanks in advance.
Eusebio.
---
Saludos/Best regards
University of Granada
Eusebio Naif Al-Soliman Fuentes MSc. Data Science and Computer Engineering
eunaif at correo.ugr.es
|