here's a cool discovery. if you have a docker container with alpine linux serving an https site, and you use busybox's wget implementation for a docker health check, you will come back to find 60% or more system cpu usage
busybox doesn't clean up the "ssl_client" processes after wget exits. i ran ps aux and it took 30 seconds to show me that there were 42,000 zombie processes belonging to the web server, from a health check at 10s intervals
atomicthumbs /
npub1q6…q45c7
2024-10-08 02:27:04
Author Public Key
npub1q6jr26pr9y8vj87a24cw25rhj5upd9d6vdtkw053vj8gywjj9y9swq45c7Published at
2024-10-08 02:27:04Event JSON
{
"id": "a48bd43197f3dde92a96fe5d8e8af2c2e7702ee64bc526dcc54127a08bd93146",
"pubkey": "06a4356823290ec91fdd5570e5507795381695ba6357673e91648e823a52290b",
"created_at": 1728354424,
"kind": 1,
"tags": [
[
"proxy",
"https://tacobelllabs.net/users/atomicthumbs/statuses/113269435550912622",
"activitypub"
]
],
"content": "here's a cool discovery. if you have a docker container with alpine linux serving an https site, and you use busybox's wget implementation for a docker health check, you will come back to find 60% or more system cpu usage\n\nbusybox doesn't clean up the \"ssl_client\" processes after wget exits. i ran ps aux and it took 30 seconds to show me that there were 42,000 zombie processes belonging to the web server, from a health check at 10s intervals",
"sig": "77207311be3ec98163302697822a94709a38f6c8e974ed010905581de1d0ba74b37106ff6ba78bb2862a7489fdc51b8f3aa73370547ca9ab28357a1bd0e5b19b"
}