Moved from GitHub flock/35
Posted by darkn3rd:
Running flock per instructions with the flock client will overload the disk and likely cause alpha to crash.
Simply, flock client running as docker containers quickly fills up the disk, possibly causing alpha to crash. If the system is still accessible, need to stop the client container, remove it, and all associated containers or images. After all links to overlay storage are removed, this space is freed up.
This to say the least is not functional for running flock tests. The flock client needs to print errors to stderr, and info to stdout. There should be a way to change the log level to only log errors, for example. If we can at least segregate stdout and stderr, there could be a way to rotate big info logs out, so that the disk does not fill up.
As per instructions.
## dgraph docker-compose up -d ## server cd go vim credentials.json docker-compose build docker-compose up -d ## client cd client docker-compose build docker-compose up -d
Flock client will generate so much traffic, the disk will be overloaded. At some point, alpha crashes (maybe due to no disk space). When flock cannot run queries, it repeatedly tries to run query, failing, and if the disk is not filled up, will definitely fill up. quickly.
I would expect the dgraph server containing the data would have the most data, but this stopped around 7-10 GB where there wasn’t any more room on 2TB drive.