Event JSON
{
"id": "c88b0e48d837d1a13fe962cce6132f0cb17f316cf93a336cb998d74836e4c878",
"pubkey": "ecf0f8f50411f90940bfa6499e9af6471ee37097e31b1b95be9bb6572b2a18fa",
"created_at": 1698060031,
"kind": 1,
"tags": [
[
"p",
"23f3b5a9df0c322c4f66590614c620cf90e2538720a106335c75239dbed152fc",
"wss://relay.mostr.pub"
],
[
"p",
"5aeb250b3075a12bd05e16c8a3c40da91a553fa92164a39915a3a0615fe51864",
"wss://relay.mostr.pub"
],
[
"e",
"58211ca5da5171aaee59e611df4c1af21de4738541f6191b2b09d6c4dcebe0ca",
"wss://relay.mostr.pub",
"reply"
],
[
"proxy",
"https://discuss.systems/users/pkhuong/statuses/111284062213271646",
"activitypub"
]
],
"content": "nostr:npub1y0emt2wlpsezcnmxtyrpf33qe7gwy5u8yzssvv6uw53em0k32t7q7smm9n Linux flushes dirty pages on a timer or when they exceed a fraction of free memory. ~10 years ago I saw spikes of badly scheduled flushing I/O with LMDB/XFS, for data that fit in RAM. You can try to tune around that's behaviour, but the failure modes when reality doesn't match the tuned values tend to be even worse (e.g., multi-minute lockups).",
"sig": "f06d239f04a906f5cbef02247543e7d3a756ac639897872b8ffd961e0306ff4655df10a45486b67d057a7fc3f1b6e7cd5590c06151aa252e5e58331f7a70cb2a"
}