I have few more queries.
1. How can I include the custom MAT plugin in JIFA?
You can provide a "hooks" class of your own. Place it on the JAR classpath for your server during startup, and then reference it in the worker configuration. More details are here:
The functions you can override are all available here
If there are more overrides needed, we can add them.
2. Can JIFA read from Object storage? ex: In case of space and memory limitation, we do not want to use JIFA as a full web application instead once we process hprof file in docker and we will
place the processed file's in object storage and delete the docker. I mean jifa_workspace path can be any object storage?
It is easiest to do this via filesystem mount point. I have implemented large object storage on the cloud in two ways (1) via S3FS, which is viable, and (2) via NFS. NFS is easier and lower latency than S3FS for sure. jifa_workspace path can then be pointed to one of these locations. But -- to do this, we use a separate system to locate and point users to the hprof filename.
We then apply a 'hook' that works out the correct NFS path for the provided path, and rewrites the path to the complete path of hprof in NFS. In mapXxxPath [1] it looks up an alternative path, check if the file exists, and provides the correct filename back for Eclipse MAT framework to read.
