Roy Dahan

So, IIUC, we don't know what's the final size of it because the backup task failed in the middle. Can you please answer my questions in this [comment](https://github.com/scylladb/scylla-manager/issues/3546#issuecomment-1718166643). Once you...

Why does the manager use /tmp? Do we have any space requirement for /tmp? @karol-kokoszka can you please look into this?

I agree it's an edge case. I thought we filled up /tmp with snapshots or manager related files. @ShlomiBalalis do we know what filled up the filesystem? How much free...

We had this issue again in https://argus.scylladb.com/test/8d7705a3-b3b8-4abd-b0e2-a0e6938c1e15/runs?additionalRuns[]=601ae9bb-6284-44b4-972e-56c816587eea. Something is filling up the root FS, hard to tell what exactly. It happened after 3 days of run and 91 nemesis...

Restoring monitor to try and check when the root FS was filled and where. From nemesis failures, I can see it happened on 4 different nodes. The newest one is...

Trying to correlate between the nemesis and free bytes in root FS (mountpoint="/"), focusing on node-1, I saw that the space drops (and never reclamied back) in 2 scenarios: 1....

Another example from node-3, same behavior. The major drop there is from "rebuild_streaming_error" that has a reboot in that node (node-3). 

Another one from node-14, drops during repair_streamin_err and mgmt_repair. 

I tried correlating it with almost every metric I could find in node_exporter, swap, files, memory, xfs, disk and couldn't find anything straight forward... One possibility it's logging, maybe one...

I don't think there is a closed matrix. @ShlomiBalalis what do we test with each version?