Intermittent internet hiccup robustness?

[stevefan1999-personal]

If we have floaty networking, or sometimes even network partitions, how do we ensure zerofs won't crash, and buffer the ongoing operations?

[Barre]

ZeroFS should properly retry reads, will stop accepting writes once the memtable is at capacity (by default 1GB), and will wait until network recovers on fsync.

[stevefan1999-personal]

Hmm, what if the network did not recover in time, will that corrupt the ZFS metadata? Even though ZFS should be robust enough for this and we have checksum and scrubs

I have to ask this because for some reason my internet may stop for a day if I had big traffic and will only recover midnight, then will keep getting grounded if I hit that certain cap again and again, kinda like an ISP level blackhole rate limiter, so I want to know if that pretty dangerous to run ZeroFS consistently.

[Barre]

I don't see why that wouldn't work, if that doesn't that would be a bug.

[tve]

I believe the OP is asking whether ZeroFS might trigger a blackholing limit in the middle of the night if the ISP resumes the connection, and I believe the answer is yes.