Syncing with full hard disk is not handled gracefully

[svarogg]

I've been syncing the mainnet blockchain to a hard-drive that apparently didn't have enough space to hold the whole blockchain.

At a certain point I saw this log:

2018-03-26 11:08:38.107 [ERR] SYNC: Failed to process block 00000000000005159e0de79c2a877a994d9a74509b2019a2051aa727b0fc44be: failed to write block to file 4 at offset 238509918: write /home/mike/.btcd/data/mainnet/blocks_ffldb/000000004.fdb: no space left on device: write /home/mike/.btcd/data/mainnet/blocks_ffldb/000000004.fdb: no space left on device

Which is good.
However, after that btcd started spamming the following logs very quickly:

2018-03-26 11:08:38.109 [INF] CHAN: Adding orphan block 00000000000006a2c0eb97ce99a2f3d749d2758feb600500b8dc6b15a4c21f79 with parent 00000000000005159e0de79c2a877a994d9a74509b2019a2051aa727b0fc44be  
2018-03-26 11:08:38.110 [INF] CHAN: Adding orphan block 00000000000000a438d46a1b52484b51cd90f4037409f6754c4943736bb7eb3d with parent 00000000000006a2c0eb97ce99a2f3d749d2758feb600500b8dc6b15a4c21f79
etc

Which isn't that good. Making it very hard to understand what the core problem was (I had to restart a couple of times and stop the moment spamming starts to find the first (real) error message).

I would expect btcd to stop sync until some hard disk space is free, or alternatively stop altogether.

I can try solving this myself if you could tell me which approach is preferred.

[abitrolly]

I remember we've run into the same issue while running btcd on Google Cloud instance. #923 (comment)