KahaDB db.redo excessive writes wearing out SSD

[Brian Lewis]

Hi, I'm investigating an issue where Islandora seems to be harassing a disk drive by doing a lot of writes, which may be wearing the drive out before its time. We're using this Docker image: https://github.com/Islandora-Collaboration-Group/ISLE

To see this issue, we did e.g. `sysdig -M60 "fd.type=file" > writes.txt` to get information about write operations to files, and then used a simple script to accumulate the number of writes and the total number of bytes written for each file.

One of the main offenders was this:

/usr/local/fedora/data/activemq-data/localhost/KahaDB/db.redo {:writes=>696, :bytes=>98976}

(this is the path inside the Docker image)

So in 60 seconds of logging, KahaDB wrote to its redo log almost 700 times. It's possible it's doing 10 writes per second around the clock. That seems excessive.

Does this issue ring a bell for anyone? Thanks.

[Noah Smith]

Hi there - that project is no longer receiving updates since the efforts of the community have almost entirely moved over to Islandora 2, but that Docker image is built from https://github.com/Islandora-Collaboration-Group/isle-fedora

I encourage you to fork that repo and change the logging settings if there's not a way to do that already with ENV variables. 

Best of luck, ~Noah