mirror of
https://github.com/neondatabase/neon.git
synced 2026-05-28 18:40:38 +00:00
Compare commits
223 Commits
problame/i
...
release-43
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
a46e77b476 | ||
|
|
a92702b01e | ||
|
|
8ff3253f20 | ||
|
|
04b82c92a7 | ||
|
|
e5bf423e68 | ||
|
|
60af392e45 | ||
|
|
661fc41e71 | ||
|
|
702c488f32 | ||
|
|
45c5122754 | ||
|
|
558394f710 | ||
|
|
73b0898608 | ||
|
|
e65be4c2dc | ||
|
|
40087b8164 | ||
|
|
c762b59483 | ||
|
|
5d71601ca9 | ||
|
|
a113c3e433 | ||
|
|
e81fc598f4 | ||
|
|
48b845fa76 | ||
|
|
27096858dc | ||
|
|
4430d0ae7d | ||
|
|
6e183aa0de | ||
|
|
fd6d0b7635 | ||
|
|
3710c32aae | ||
|
|
be83bee49d | ||
|
|
cf28e5922a | ||
|
|
7d384d6953 | ||
|
|
4b3b37b912 | ||
|
|
1d8d200f4d | ||
|
|
0d80d6ce18 | ||
|
|
f653ee039f | ||
|
|
e614a95853 | ||
|
|
850db4cc13 | ||
|
|
8a316b1277 | ||
|
|
4d13bae449 | ||
|
|
49377abd98 | ||
|
|
a6b2f4e54e | ||
|
|
face60d50b | ||
|
|
9768aa27f2 | ||
|
|
96b2e575e1 | ||
|
|
7222777784 | ||
|
|
5469fdede0 | ||
|
|
72aa6b9fdd | ||
|
|
ae0634b7be | ||
|
|
70711f32fa | ||
|
|
52a88af0aa | ||
|
|
b7a43bf817 | ||
|
|
dce91b33a4 | ||
|
|
23ee4f3050 | ||
|
|
46857e8282 | ||
|
|
368ab0ce54 | ||
|
|
a5987eebfd | ||
|
|
6686ede30f | ||
|
|
373c7057cc | ||
|
|
7d6ec16166 | ||
|
|
0e6fdc8a58 | ||
|
|
521438a5c6 | ||
|
|
07d7874bc8 | ||
|
|
1804111a02 | ||
|
|
cd0178efed | ||
|
|
333574be57 | ||
|
|
79a799a143 | ||
|
|
9da06af6c9 | ||
|
|
ce1753d036 | ||
|
|
67db8432b4 | ||
|
|
4e2e44e524 | ||
|
|
ed786104f3 | ||
|
|
84b74f2bd1 | ||
|
|
fec2ad6283 | ||
|
|
98eebd4682 | ||
|
|
2f74287c9b | ||
|
|
aee1bf95e3 | ||
|
|
b9de9d75ff | ||
|
|
7943b709e6 | ||
|
|
d7d066d493 | ||
|
|
e78ac22107 | ||
|
|
76a8f2bb44 | ||
|
|
8d59a8581f | ||
|
|
b1ddd01289 | ||
|
|
6eae4fc9aa | ||
|
|
765455bca2 | ||
|
|
4204960942 | ||
|
|
67345d66ea | ||
|
|
2266ee5971 | ||
|
|
b58445d855 | ||
|
|
36050e7f3d | ||
|
|
33360ed96d | ||
|
|
39a28d1108 | ||
|
|
efa6aa134f | ||
|
|
2c724e56e2 | ||
|
|
feff887c6f | ||
|
|
353d915fcf | ||
|
|
2e38098cbc | ||
|
|
a6fe5ea1ac | ||
|
|
05b0aed0c1 | ||
|
|
cd1705357d | ||
|
|
6bc7561290 | ||
|
|
fbd3ac14b5 | ||
|
|
e437787c8f | ||
|
|
3460dbf90b | ||
|
|
6b89d99677 | ||
|
|
6cc8ea86e4 | ||
|
|
e62a492d6f | ||
|
|
a475cdf642 | ||
|
|
7002c79a47 | ||
|
|
ee6cf357b4 | ||
|
|
e5c2086b5f | ||
|
|
5f1208296a | ||
|
|
88e8e473cd | ||
|
|
b0a77844f6 | ||
|
|
1baf464307 | ||
|
|
e9b8e81cea | ||
|
|
85d6194aa4 | ||
|
|
333a7a68ef | ||
|
|
6aa4e41bee | ||
|
|
840183e51f | ||
|
|
cbccc94b03 | ||
|
|
fce227df22 | ||
|
|
bd787e800f | ||
|
|
4a7704b4a3 | ||
|
|
ff1119da66 | ||
|
|
4c3ba1627b | ||
|
|
1407174fb2 | ||
|
|
ec9dcb1889 | ||
|
|
d11d781afc | ||
|
|
4e44565b71 | ||
|
|
4ed51ad33b | ||
|
|
1c1ebe5537 | ||
|
|
c19cb7f386 | ||
|
|
4b97d31b16 | ||
|
|
923ade3dd7 | ||
|
|
b04e711975 | ||
|
|
afd0a6b39a | ||
|
|
99752286d8 | ||
|
|
15df93363c | ||
|
|
bc0ab741af | ||
|
|
51d9dfeaa3 | ||
|
|
f63cb18155 | ||
|
|
0de603d88e | ||
|
|
240913912a | ||
|
|
91a4ea0de2 | ||
|
|
8608704f49 | ||
|
|
efef68ce99 | ||
|
|
8daefd24da | ||
|
|
46cc8b7982 | ||
|
|
38cd90dd0c | ||
|
|
a51b269f15 | ||
|
|
43bf6d0a0f | ||
|
|
15273a9b66 | ||
|
|
78aca668d0 | ||
|
|
acbf4148ea | ||
|
|
6508540561 | ||
|
|
a41b5244a8 | ||
|
|
2b3189be95 | ||
|
|
248563c595 | ||
|
|
14cd6ca933 | ||
|
|
eb36403e71 | ||
|
|
3c6f779698 | ||
|
|
f67f0c1c11 | ||
|
|
edb02d3299 | ||
|
|
664a69e65b | ||
|
|
478322ebf9 | ||
|
|
802f174072 | ||
|
|
47f9890bae | ||
|
|
262265daad | ||
|
|
300da5b872 | ||
|
|
7b22b5c433 | ||
|
|
ffca97bc1e | ||
|
|
cb356f3259 | ||
|
|
c85374295f | ||
|
|
4992160677 | ||
|
|
bd535b3371 | ||
|
|
d90c5a03af | ||
|
|
2d02cc9079 | ||
|
|
49ad94b99f | ||
|
|
948a217398 | ||
|
|
125381eae7 | ||
|
|
cd01bbc715 | ||
|
|
d8b5e3b88d | ||
|
|
06d25f2186 | ||
|
|
f759b561f3 | ||
|
|
ece0555600 | ||
|
|
73ea0a0b01 | ||
|
|
d8f6d6fd6f | ||
|
|
d24de169a7 | ||
|
|
0816168296 | ||
|
|
277b44d57a | ||
|
|
68c2c3880e | ||
|
|
49da498f65 | ||
|
|
2c76ba3dd7 | ||
|
|
dbe3dc69ad | ||
|
|
8e5bb3ed49 | ||
|
|
ab0be7b8da | ||
|
|
b4c55f5d24 | ||
|
|
ede70d833c | ||
|
|
70c3d18bb0 | ||
|
|
7a491f52c4 | ||
|
|
323c4ecb4f | ||
|
|
3d2466607e | ||
|
|
ed478b39f4 | ||
|
|
91585a558d | ||
|
|
93467eae1f | ||
|
|
f3aac81d19 | ||
|
|
979ad60c19 | ||
|
|
9316cb1b1f | ||
|
|
e7939a527a | ||
|
|
36d26665e1 | ||
|
|
873347f977 | ||
|
|
e814ac16f9 | ||
|
|
ad3055d386 | ||
|
|
94e03eb452 | ||
|
|
380f26ef79 | ||
|
|
3c5b7f59d7 | ||
|
|
fee89f80b5 | ||
|
|
41cce8eaf1 | ||
|
|
f88fe0218d | ||
|
|
cc856eca85 | ||
|
|
cf350c6002 | ||
|
|
0ce6b6a0a3 | ||
|
|
73f247d537 | ||
|
|
960be82183 | ||
|
|
806e5a6c19 | ||
|
|
8d5df07cce | ||
|
|
df7a9d1407 |
8
.github/workflows/build_and_test.yml
vendored
8
.github/workflows/build_and_test.yml
vendored
@@ -199,6 +199,10 @@ jobs:
|
||||
#
|
||||
git config --global --add safe.directory ${{ github.workspace }}
|
||||
git config --global --add safe.directory ${GITHUB_WORKSPACE}
|
||||
for r in 14 15 16; do
|
||||
git config --global --add safe.directory "${{ github.workspace }}/vendor/postgres-v$r"
|
||||
git config --global --add safe.directory "${GITHUB_WORKSPACE}/vendor/postgres-v$r"
|
||||
done
|
||||
|
||||
- name: Checkout
|
||||
uses: actions/checkout@v3
|
||||
@@ -1097,6 +1101,10 @@ jobs:
|
||||
#
|
||||
git config --global --add safe.directory ${{ github.workspace }}
|
||||
git config --global --add safe.directory ${GITHUB_WORKSPACE}
|
||||
for r in 14 15 16; do
|
||||
git config --global --add safe.directory "${{ github.workspace }}/vendor/postgres-v$r"
|
||||
git config --global --add safe.directory "${GITHUB_WORKSPACE}/vendor/postgres-v$r"
|
||||
done
|
||||
|
||||
- name: Checkout
|
||||
uses: actions/checkout@v3
|
||||
|
||||
18
.github/workflows/neon_extra_builds.yml
vendored
18
.github/workflows/neon_extra_builds.yml
vendored
@@ -142,6 +142,10 @@ jobs:
|
||||
#
|
||||
git config --global --add safe.directory ${{ github.workspace }}
|
||||
git config --global --add safe.directory ${GITHUB_WORKSPACE}
|
||||
for r in 14 15 16; do
|
||||
git config --global --add safe.directory "${{ github.workspace }}/vendor/postgres-v$r"
|
||||
git config --global --add safe.directory "${GITHUB_WORKSPACE}/vendor/postgres-v$r"
|
||||
done
|
||||
|
||||
- name: Checkout
|
||||
uses: actions/checkout@v4
|
||||
@@ -238,6 +242,20 @@ jobs:
|
||||
options: --init
|
||||
|
||||
steps:
|
||||
- name: Fix git ownership
|
||||
run: |
|
||||
# Workaround for `fatal: detected dubious ownership in repository at ...`
|
||||
#
|
||||
# Use both ${{ github.workspace }} and ${GITHUB_WORKSPACE} because they're different on host and in containers
|
||||
# Ref https://github.com/actions/checkout/issues/785
|
||||
#
|
||||
git config --global --add safe.directory ${{ github.workspace }}
|
||||
git config --global --add safe.directory ${GITHUB_WORKSPACE}
|
||||
for r in 14 15 16; do
|
||||
git config --global --add safe.directory "${{ github.workspace }}/vendor/postgres-v$r"
|
||||
git config --global --add safe.directory "${GITHUB_WORKSPACE}/vendor/postgres-v$r"
|
||||
done
|
||||
|
||||
- name: Checkout
|
||||
uses: actions/checkout@v4
|
||||
with:
|
||||
|
||||
95
Cargo.lock
generated
95
Cargo.lock
generated
@@ -184,7 +184,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "81953c529336010edd6d8e358f886d9581267795c61b19475b71314bffa46d35"
|
||||
dependencies = [
|
||||
"concurrent-queue",
|
||||
"event-listener 2.5.3",
|
||||
"event-listener",
|
||||
"futures-core",
|
||||
]
|
||||
|
||||
@@ -205,13 +205,11 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "async-lock"
|
||||
version = "3.2.0"
|
||||
version = "2.8.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "7125e42787d53db9dd54261812ef17e937c95a51e4d291373b670342fa44310c"
|
||||
checksum = "287272293e9d8c41773cec55e365490fe034813a2f172f502d6ddcf75b2f582b"
|
||||
dependencies = [
|
||||
"event-listener 4.0.0",
|
||||
"event-listener-strategy",
|
||||
"pin-project-lite",
|
||||
"event-listener",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
@@ -694,9 +692,9 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "azure_core"
|
||||
version = "0.17.0"
|
||||
version = "0.16.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "4ccd63c07d1fbfb3d4543d7ea800941bf5a30db1911b9b9e4db3b2c4210a434f"
|
||||
checksum = "8e29286b9edfdd6f2c7e9d970bb5b015df8621258acab9ecfcea09b2d7692467"
|
||||
dependencies = [
|
||||
"async-trait",
|
||||
"base64 0.21.1",
|
||||
@@ -721,9 +719,9 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "azure_identity"
|
||||
version = "0.17.0"
|
||||
version = "0.16.2"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "8bd7ea32ca7eb66ff4757f83baac702ff11d469e5de365b6bc6f79f9c25d3436"
|
||||
checksum = "5b67b337346da8739e91ea1e9400a6ebc9bc54e0b2af1d23c9bcd565950588f9"
|
||||
dependencies = [
|
||||
"async-lock",
|
||||
"async-trait",
|
||||
@@ -742,9 +740,9 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "azure_storage"
|
||||
version = "0.17.0"
|
||||
version = "0.16.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "83ca0a07f89fd72a006da4713e93af3d6c44a693e61a1c3c2e7985de39c182e8"
|
||||
checksum = "bed0ccefde57930b2886fd4aed1f70ac469c197b8c2e94828290d71bcbdb5d97"
|
||||
dependencies = [
|
||||
"RustyXML",
|
||||
"async-trait",
|
||||
@@ -764,9 +762,9 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "azure_storage_blobs"
|
||||
version = "0.17.0"
|
||||
version = "0.16.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "8096c04d370118323c42b2752aa1883e4880a56ef65239f317b359f263b6e194"
|
||||
checksum = "f91a52da2d192cfe43759f61e8bb31a5969f1722d5b85ac89627f356ad674ab4"
|
||||
dependencies = [
|
||||
"RustyXML",
|
||||
"azure_core",
|
||||
@@ -1688,27 +1686,6 @@ version = "2.5.3"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "0206175f82b8d6bf6652ff7d71a1e27fd2e4efde587fd368662814d6ec1d9ce0"
|
||||
|
||||
[[package]]
|
||||
name = "event-listener"
|
||||
version = "4.0.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "770d968249b5d99410d61f5bf89057f3199a077a04d087092f58e7d10692baae"
|
||||
dependencies = [
|
||||
"concurrent-queue",
|
||||
"parking",
|
||||
"pin-project-lite",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "event-listener-strategy"
|
||||
version = "0.4.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "958e4d70b6d5e81971bebec42271ec641e7ff4e170a6fa605f2b8a8b65cb97d3"
|
||||
dependencies = [
|
||||
"event-listener 4.0.0",
|
||||
"pin-project-lite",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "fail"
|
||||
version = "0.5.1"
|
||||
@@ -2413,16 +2390,6 @@ dependencies = [
|
||||
"windows-sys 0.48.0",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "io-uring"
|
||||
version = "0.6.2"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "460648e47a07a43110fbfa2e0b14afb2be920093c31e5dccc50e49568e099762"
|
||||
dependencies = [
|
||||
"bitflags 1.3.2",
|
||||
"libc",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "ipnet"
|
||||
version = "2.9.0"
|
||||
@@ -3131,7 +3098,6 @@ dependencies = [
|
||||
"tenant_size_model",
|
||||
"thiserror",
|
||||
"tokio",
|
||||
"tokio-epoll-uring",
|
||||
"tokio-io-timeout",
|
||||
"tokio-postgres",
|
||||
"tokio-tar",
|
||||
@@ -3712,9 +3678,9 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "quick-xml"
|
||||
version = "0.31.0"
|
||||
version = "0.30.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "1004a344b30a54e2ee58d66a71b32d2db2feb0a31f9a2d302bf0536f15de2a33"
|
||||
checksum = "eff6510e86862b57b210fd8cbe8ed3f0d7d600b9c2863cd4549a2e033c66e956"
|
||||
dependencies = [
|
||||
"memchr",
|
||||
"serde",
|
||||
@@ -5039,18 +5005,18 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "thiserror"
|
||||
version = "1.0.47"
|
||||
version = "1.0.40"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "97a802ec30afc17eee47b2855fc72e0c4cd62be9b4efe6591edde0ec5bd68d8f"
|
||||
checksum = "978c9a314bd8dc99be594bc3c175faaa9794be04a5a5e153caba6915336cebac"
|
||||
dependencies = [
|
||||
"thiserror-impl",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "thiserror-impl"
|
||||
version = "1.0.47"
|
||||
version = "1.0.40"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "6bb623b56e39ab7dcd4b1b98bb6c8f8d907ed255b18de254088016b27a8ee19b"
|
||||
checksum = "f9456a42c5b0d803c8cd86e73dd7cc9edd429499f37a3550d286d5e86720569f"
|
||||
dependencies = [
|
||||
"proc-macro2",
|
||||
"quote",
|
||||
@@ -5147,7 +5113,6 @@ dependencies = [
|
||||
"libc",
|
||||
"mio",
|
||||
"num_cpus",
|
||||
"parking_lot 0.12.1",
|
||||
"pin-project-lite",
|
||||
"signal-hook-registry",
|
||||
"socket2 0.5.5",
|
||||
@@ -5155,21 +5120,6 @@ dependencies = [
|
||||
"windows-sys 0.48.0",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "tokio-epoll-uring"
|
||||
version = "0.1.0"
|
||||
source = "git+https://github.com/neondatabase/tokio-epoll-uring.git?branch=main#2ae71cebd8f032ebfa987b713a77bc3d23aac57c"
|
||||
dependencies = [
|
||||
"futures",
|
||||
"once_cell",
|
||||
"scopeguard",
|
||||
"thiserror",
|
||||
"tokio",
|
||||
"tokio-util",
|
||||
"tracing",
|
||||
"uring-common",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "tokio-io-timeout"
|
||||
version = "1.2.0"
|
||||
@@ -5711,15 +5661,6 @@ dependencies = [
|
||||
"webpki-roots 0.23.1",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "uring-common"
|
||||
version = "0.1.0"
|
||||
source = "git+https://github.com/neondatabase/tokio-epoll-uring.git?branch=main#2ae71cebd8f032ebfa987b713a77bc3d23aac57c"
|
||||
dependencies = [
|
||||
"io-uring",
|
||||
"libc",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "url"
|
||||
version = "2.3.1"
|
||||
|
||||
@@ -38,10 +38,10 @@ license = "Apache-2.0"
|
||||
anyhow = { version = "1.0", features = ["backtrace"] }
|
||||
arc-swap = "1.6"
|
||||
async-compression = { version = "0.4.0", features = ["tokio", "gzip", "zstd"] }
|
||||
azure_core = "0.17"
|
||||
azure_identity = "0.17"
|
||||
azure_storage = "0.17"
|
||||
azure_storage_blobs = "0.17"
|
||||
azure_core = "0.16"
|
||||
azure_identity = "0.16"
|
||||
azure_storage = "0.16"
|
||||
azure_storage_blobs = "0.16"
|
||||
flate2 = "1.0.26"
|
||||
async-stream = "0.3"
|
||||
async-trait = "0.1"
|
||||
|
||||
@@ -83,8 +83,6 @@ enum-map.workspace = true
|
||||
enumset.workspace = true
|
||||
strum.workspace = true
|
||||
strum_macros.workspace = true
|
||||
#tokio-epoll-uring = { path = "../../tokio-epoll-uring/tokio-epoll-uring" }
|
||||
tokio-epoll-uring = { git = "https://github.com/neondatabase/tokio-epoll-uring.git" , branch = "main" }
|
||||
|
||||
[dev-dependencies]
|
||||
criterion.workspace = true
|
||||
|
||||
@@ -281,12 +281,18 @@ async fn calculate_synthetic_size_worker(
|
||||
// By using the same limiter, we centralize metrics collection for "start" and "finished" counters,
|
||||
// which turns out is really handy to understand the system.
|
||||
if let Err(e) = tenant.calculate_synthetic_size(cause, cancel, ctx).await {
|
||||
if let Some(PageReconstructError::Cancelled) =
|
||||
e.downcast_ref::<PageReconstructError>()
|
||||
{
|
||||
return Ok(());
|
||||
// this error can be returned if timeline is shutting down, but it does not
|
||||
// mean the synthetic size worker should terminate. we do not need any checks
|
||||
// in this function because `mgr::get_tenant` will error out after shutdown has
|
||||
// progressed to shutting down tenants.
|
||||
let is_cancelled = matches!(
|
||||
e.downcast_ref::<PageReconstructError>(),
|
||||
Some(PageReconstructError::Cancelled)
|
||||
);
|
||||
|
||||
if !is_cancelled {
|
||||
error!("failed to calculate synthetic size for tenant {tenant_id}: {e:#}");
|
||||
}
|
||||
error!("failed to calculate synthetic size for tenant {tenant_id}: {e:#}");
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -299,7 +305,7 @@ async fn calculate_synthetic_size_worker(
|
||||
|
||||
let res = tokio::time::timeout_at(
|
||||
started_at + synthetic_size_calculation_interval,
|
||||
task_mgr::shutdown_token().cancelled(),
|
||||
cancel.cancelled(),
|
||||
)
|
||||
.await;
|
||||
if res.is_ok() {
|
||||
|
||||
@@ -5,10 +5,10 @@
|
||||
use super::ephemeral_file::EphemeralFile;
|
||||
use super::storage_layer::delta_layer::{Adapter, DeltaLayerInner};
|
||||
use crate::context::RequestContext;
|
||||
use crate::page_cache::{self, PageReadGuard, PageWriteGuard, ReadBufResult, PAGE_SZ};
|
||||
use crate::page_cache::{self, PageReadGuard, ReadBufResult, PAGE_SZ};
|
||||
use crate::virtual_file::VirtualFile;
|
||||
use bytes::Bytes;
|
||||
use std::ops::Deref;
|
||||
use std::ops::{Deref, DerefMut};
|
||||
|
||||
/// This is implemented by anything that can read 8 kB (PAGE_SZ)
|
||||
/// blocks, using the page cache
|
||||
@@ -39,8 +39,6 @@ pub enum BlockLease<'a> {
|
||||
EphemeralFileMutableTail(&'a [u8; PAGE_SZ]),
|
||||
#[cfg(test)]
|
||||
Arc(std::sync::Arc<[u8; PAGE_SZ]>),
|
||||
#[cfg(test)]
|
||||
Vec(Vec<u8>),
|
||||
}
|
||||
|
||||
impl From<PageReadGuard<'static>> for BlockLease<'static> {
|
||||
@@ -65,13 +63,6 @@ impl<'a> Deref for BlockLease<'a> {
|
||||
BlockLease::EphemeralFileMutableTail(v) => v,
|
||||
#[cfg(test)]
|
||||
BlockLease::Arc(v) => v.deref(),
|
||||
#[cfg(test)]
|
||||
BlockLease::Vec(v) => {
|
||||
let v: &Vec<u8> = v;
|
||||
assert_eq!(v.len(), PAGE_SZ, "caller must ensure that v has PAGE_SZ");
|
||||
// Safety: see above assertion.
|
||||
unsafe { &*(v.as_ptr() as *const [u8; PAGE_SZ]) }
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -178,14 +169,10 @@ impl FileBlockReader {
|
||||
}
|
||||
|
||||
/// Read a page from the underlying file into given buffer.
|
||||
async fn fill_buffer(
|
||||
&self,
|
||||
buf: PageWriteGuard<'static>,
|
||||
blkno: u32,
|
||||
) -> Result<PageWriteGuard<'static>, std::io::Error> {
|
||||
async fn fill_buffer(&self, buf: &mut [u8], blkno: u32) -> Result<(), std::io::Error> {
|
||||
assert!(buf.len() == PAGE_SZ);
|
||||
self.file
|
||||
.read_exact_at_page(buf, blkno as u64 * PAGE_SZ as u64)
|
||||
.read_exact_at(buf, blkno as u64 * PAGE_SZ as u64)
|
||||
.await
|
||||
}
|
||||
/// Read a block.
|
||||
@@ -209,9 +196,9 @@ impl FileBlockReader {
|
||||
)
|
||||
})? {
|
||||
ReadBufResult::Found(guard) => Ok(guard.into()),
|
||||
ReadBufResult::NotFound(write_guard) => {
|
||||
ReadBufResult::NotFound(mut write_guard) => {
|
||||
// Read the page from disk into the buffer
|
||||
let write_guard = self.fill_buffer(write_guard, blknum).await?;
|
||||
self.fill_buffer(write_guard.deref_mut(), blknum).await?;
|
||||
Ok(write_guard.mark_valid().into())
|
||||
}
|
||||
}
|
||||
|
||||
@@ -9,6 +9,7 @@ use crate::virtual_file::VirtualFile;
|
||||
use camino::Utf8PathBuf;
|
||||
use pageserver_api::shard::TenantShardId;
|
||||
use std::cmp::min;
|
||||
use std::fs::OpenOptions;
|
||||
use std::io::{self, ErrorKind};
|
||||
use std::ops::DerefMut;
|
||||
use std::sync::atomic::AtomicU64;
|
||||
@@ -44,11 +45,11 @@ impl EphemeralFile {
|
||||
"ephemeral-{filename_disambiguator}"
|
||||
)));
|
||||
|
||||
let file = {
|
||||
let mut options = tokio_epoll_uring::ops::open_at::OpenOptions::new();
|
||||
options.read(true).write(true).create(true);
|
||||
VirtualFile::open_with_options_async(&filename, options).await?
|
||||
};
|
||||
let file = VirtualFile::open_with_options(
|
||||
&filename,
|
||||
OpenOptions::new().read(true).write(true).create(true),
|
||||
)
|
||||
.await?;
|
||||
|
||||
Ok(EphemeralFile {
|
||||
page_cache_file_id: page_cache::next_file_id(),
|
||||
@@ -88,10 +89,11 @@ impl EphemeralFile {
|
||||
page_cache::ReadBufResult::Found(guard) => {
|
||||
return Ok(BlockLease::PageReadGuard(guard))
|
||||
}
|
||||
page_cache::ReadBufResult::NotFound(write_guard) => {
|
||||
let write_guard = self
|
||||
.file
|
||||
.read_exact_at_page(write_guard, blknum as u64 * PAGE_SZ as u64)
|
||||
page_cache::ReadBufResult::NotFound(mut write_guard) => {
|
||||
let buf: &mut [u8] = write_guard.deref_mut();
|
||||
debug_assert_eq!(buf.len(), PAGE_SZ);
|
||||
self.file
|
||||
.read_exact_at(&mut buf[..], blknum as u64 * PAGE_SZ as u64)
|
||||
.await?;
|
||||
let read_guard = write_guard.mark_valid();
|
||||
return Ok(BlockLease::PageReadGuard(read_guard));
|
||||
|
||||
@@ -647,13 +647,12 @@ impl DeltaLayer {
|
||||
where
|
||||
F: Fn(Summary) -> Summary,
|
||||
{
|
||||
let file = {
|
||||
let mut options = tokio_epoll_uring::ops::open_at::OpenOptions::new();
|
||||
options.read(true).write(true);
|
||||
VirtualFile::open_with_options_async(path, options)
|
||||
.await
|
||||
.with_context(|| format!("Failed to open file '{}'", path))?
|
||||
};
|
||||
let file = VirtualFile::open_with_options(
|
||||
path,
|
||||
&*std::fs::OpenOptions::new().read(true).write(true),
|
||||
)
|
||||
.await
|
||||
.with_context(|| format!("Failed to open file '{}'", path))?;
|
||||
let file = FileBlockReader::new(file);
|
||||
let summary_blk = file.read_blk(0, ctx).await?;
|
||||
let actual_summary = Summary::des_prefix(summary_blk.as_ref()).context("deserialize")?;
|
||||
|
||||
@@ -325,13 +325,12 @@ impl ImageLayer {
|
||||
where
|
||||
F: Fn(Summary) -> Summary,
|
||||
{
|
||||
let file = {
|
||||
let mut options = tokio_epoll_uring::ops::open_at::OpenOptions::new();
|
||||
options.read(true).write(true);
|
||||
VirtualFile::open_with_options_async(path, options)
|
||||
.await
|
||||
.with_context(|| format!("Failed to open file '{}'", path))?
|
||||
};
|
||||
let file = VirtualFile::open_with_options(
|
||||
path,
|
||||
&*std::fs::OpenOptions::new().read(true).write(true),
|
||||
)
|
||||
.await
|
||||
.with_context(|| format!("Failed to open file '{}'", path))?;
|
||||
let file = FileBlockReader::new(file);
|
||||
let summary_blk = file.read_blk(0, ctx).await?;
|
||||
let actual_summary = Summary::des_prefix(summary_blk.as_ref()).context("deserialize")?;
|
||||
@@ -493,11 +492,11 @@ impl ImageLayerWriterInner {
|
||||
},
|
||||
);
|
||||
info!("new image layer {path}");
|
||||
let mut file = {
|
||||
let mut options = tokio_epoll_uring::ops::open_at::OpenOptions::new();
|
||||
options.write(true).create_new(true);
|
||||
VirtualFile::open_with_options_async(&path, options).await?
|
||||
};
|
||||
let mut file = VirtualFile::open_with_options(
|
||||
&path,
|
||||
std::fs::OpenOptions::new().write(true).create_new(true),
|
||||
)
|
||||
.await?;
|
||||
// make room for the header block
|
||||
file.seek(SeekFrom::Start(PAGE_SZ as u64)).await?;
|
||||
let blob_writer = BlobWriter::new(file, PAGE_SZ as u64);
|
||||
|
||||
@@ -11,17 +11,14 @@
|
||||
//! src/backend/storage/file/fd.c
|
||||
//!
|
||||
use crate::metrics::{StorageIoOperation, STORAGE_IO_SIZE, STORAGE_IO_TIME_METRIC};
|
||||
use crate::page_cache::PageWriteGuard;
|
||||
use crate::tenant::TENANTS_SEGMENT_NAME;
|
||||
use camino::{Utf8Path, Utf8PathBuf};
|
||||
use once_cell::sync::OnceCell;
|
||||
use std::fs::{self, File};
|
||||
use std::fs::{self, File, OpenOptions};
|
||||
use std::io::{Error, ErrorKind, Seek, SeekFrom};
|
||||
use std::os::fd::{AsRawFd, FromRawFd, IntoRawFd, OwnedFd};
|
||||
use std::os::unix::fs::FileExt;
|
||||
use std::sync::atomic::{AtomicBool, AtomicUsize, Ordering};
|
||||
use tokio::sync::{RwLock, RwLockReadGuard, RwLockWriteGuard};
|
||||
use tokio::time::Instant;
|
||||
use tokio_epoll_uring::IoBufMut;
|
||||
use std::sync::{RwLock, RwLockWriteGuard};
|
||||
use utils::fs_ext;
|
||||
|
||||
///
|
||||
@@ -56,7 +53,7 @@ pub struct VirtualFile {
|
||||
/// opened, in the VirtualFile::create() function, and strip the flag before
|
||||
/// storing it here.
|
||||
pub path: Utf8PathBuf,
|
||||
open_options: tokio_epoll_uring::ops::open_at::OpenOptions,
|
||||
open_options: OpenOptions,
|
||||
|
||||
// These are strings becase we only use them for metrics, and those expect strings.
|
||||
// It makes no sense for us to constantly turn the `TimelineId` and `TenantId` into
|
||||
@@ -109,43 +106,12 @@ struct SlotInner {
|
||||
file: Option<File>,
|
||||
}
|
||||
|
||||
/// Impl of [`tokio_epoll_uring::IoBuf`] and [`tokio_epoll_uring::IoBufMut`] for [`PageWriteGuard`].
|
||||
struct PageWriteGuardBuf {
|
||||
page: PageWriteGuard<'static>,
|
||||
init_up_to: usize,
|
||||
}
|
||||
// Safety: the [`PageWriteGuard`] gives us exclusive ownership of the page cache slot,
|
||||
// and the location remains stable even if [`Self`] or the [`PageWriteGuard`] is moved.
|
||||
unsafe impl tokio_epoll_uring::IoBuf for PageWriteGuardBuf {
|
||||
fn stable_ptr(&self) -> *const u8 {
|
||||
self.page.as_ptr()
|
||||
}
|
||||
fn bytes_init(&self) -> usize {
|
||||
self.init_up_to
|
||||
}
|
||||
fn bytes_total(&self) -> usize {
|
||||
self.page.len()
|
||||
}
|
||||
}
|
||||
// Safety: see above, plus: the ownership of [`PageWriteGuard`] means exclusive access,
|
||||
// hence it's safe to hand out the `stable_mut_ptr()`.
|
||||
unsafe impl tokio_epoll_uring::IoBufMut for PageWriteGuardBuf {
|
||||
fn stable_mut_ptr(&mut self) -> *mut u8 {
|
||||
self.page.as_mut_ptr()
|
||||
}
|
||||
|
||||
unsafe fn set_init(&mut self, pos: usize) {
|
||||
assert!(pos <= self.page.len());
|
||||
self.init_up_to = pos;
|
||||
}
|
||||
}
|
||||
|
||||
impl OpenFiles {
|
||||
/// Find a slot to use, evicting an existing file descriptor if needed.
|
||||
///
|
||||
/// On return, we hold a lock on the slot, and its 'tag' has been updated
|
||||
/// recently_used has been set. It's all ready for reuse.
|
||||
async fn find_victim_slot(&self) -> (SlotHandle, RwLockWriteGuard<SlotInner>) {
|
||||
fn find_victim_slot(&self) -> (SlotHandle, RwLockWriteGuard<SlotInner>) {
|
||||
//
|
||||
// Run the clock algorithm to find a slot to replace.
|
||||
//
|
||||
@@ -177,7 +143,7 @@ impl OpenFiles {
|
||||
}
|
||||
retries += 1;
|
||||
} else {
|
||||
slot_guard = slot.inner.write().await;
|
||||
slot_guard = slot.inner.write().unwrap();
|
||||
index = next;
|
||||
break;
|
||||
}
|
||||
@@ -188,7 +154,7 @@ impl OpenFiles {
|
||||
// old file.
|
||||
//
|
||||
if let Some(old_file) = slot_guard.file.take() {
|
||||
// the normal path of dropping VirtualFile uses `Close`, use `CloseByReplace` here to
|
||||
// the normal path of dropping VirtualFile uses "close", use "close-by-replace" here to
|
||||
// distinguish the two.
|
||||
STORAGE_IO_TIME_METRIC
|
||||
.get(StorageIoOperation::CloseByReplace)
|
||||
@@ -284,43 +250,76 @@ impl<T> MaybeFatalIo<T> for std::io::Result<T> {
|
||||
}
|
||||
}
|
||||
|
||||
/// Observe duration for the given storage I/O operation
|
||||
///
|
||||
/// Unlike `observe_closure_duration`, this supports async,
|
||||
/// where "support" means that we measure wall clock time.
|
||||
macro_rules! observe_duration {
|
||||
($op:expr, $($body:tt)*) => {{
|
||||
let instant = Instant::now();
|
||||
let result = $($body)*;
|
||||
let elapsed = instant.elapsed().as_secs_f64();
|
||||
STORAGE_IO_TIME_METRIC
|
||||
.get($op)
|
||||
.observe(elapsed);
|
||||
result
|
||||
}}
|
||||
}
|
||||
|
||||
macro_rules! with_file {
|
||||
($this:expr, $op:expr, | $ident:ident | $($body:tt)*) => {{
|
||||
let $ident = $this.lock_file().await?;
|
||||
observe_duration!($op, $($body)*)
|
||||
}};
|
||||
}
|
||||
|
||||
impl VirtualFile {
|
||||
/// Open a file in read-only mode. Like File::open.
|
||||
pub async fn open(path: &Utf8Path) -> Result<VirtualFile, std::io::Error> {
|
||||
let mut options = tokio_epoll_uring::ops::open_at::OpenOptions::new();
|
||||
options.read(true);
|
||||
Self::open_with_options_async(path, options).await
|
||||
Self::open_with_options(path, OpenOptions::new().read(true)).await
|
||||
}
|
||||
|
||||
/// Create a new file for writing. If the file exists, it will be truncated.
|
||||
/// Like File::create.
|
||||
pub async fn create(path: &Utf8Path) -> Result<VirtualFile, std::io::Error> {
|
||||
let mut options = tokio_epoll_uring::ops::open_at::OpenOptions::new();
|
||||
options.write(true).create(true).truncate(true);
|
||||
Self::open_with_options_async(path, options).await
|
||||
Self::open_with_options(
|
||||
path,
|
||||
OpenOptions::new().write(true).create(true).truncate(true),
|
||||
)
|
||||
.await
|
||||
}
|
||||
|
||||
/// Open a file with given options.
|
||||
///
|
||||
/// Note: If any custom flags were set in 'open_options' through OpenOptionsExt,
|
||||
/// they will be applied also when the file is subsequently re-opened, not only
|
||||
/// on the first time. Make sure that's sane!
|
||||
pub async fn open_with_options(
|
||||
path: &Utf8Path,
|
||||
open_options: &OpenOptions,
|
||||
) -> Result<VirtualFile, std::io::Error> {
|
||||
let path_str = path.to_string();
|
||||
let parts = path_str.split('/').collect::<Vec<&str>>();
|
||||
let tenant_id;
|
||||
let timeline_id;
|
||||
if parts.len() > 5 && parts[parts.len() - 5] == TENANTS_SEGMENT_NAME {
|
||||
tenant_id = parts[parts.len() - 4].to_string();
|
||||
timeline_id = parts[parts.len() - 2].to_string();
|
||||
} else {
|
||||
tenant_id = "*".to_string();
|
||||
timeline_id = "*".to_string();
|
||||
}
|
||||
let (handle, mut slot_guard) = get_open_files().find_victim_slot();
|
||||
|
||||
// NB: there is also StorageIoOperation::OpenAfterReplace which is for the case
|
||||
// where our caller doesn't get to use the returned VirtualFile before its
|
||||
// slot gets re-used by someone else.
|
||||
let file = STORAGE_IO_TIME_METRIC
|
||||
.get(StorageIoOperation::Open)
|
||||
.observe_closure_duration(|| open_options.open(path))?;
|
||||
|
||||
// Strip all options other than read and write.
|
||||
//
|
||||
// It would perhaps be nicer to check just for the read and write flags
|
||||
// explicitly, but OpenOptions doesn't contain any functions to read flags,
|
||||
// only to set them.
|
||||
let mut reopen_options = open_options.clone();
|
||||
reopen_options.create(false);
|
||||
reopen_options.create_new(false);
|
||||
reopen_options.truncate(false);
|
||||
|
||||
let vfile = VirtualFile {
|
||||
handle: RwLock::new(handle),
|
||||
pos: 0,
|
||||
path: path.to_path_buf(),
|
||||
open_options: reopen_options,
|
||||
tenant_id,
|
||||
timeline_id,
|
||||
};
|
||||
|
||||
// TODO: Under pressure, it's likely the slot will get re-used and
|
||||
// the underlying file closed before they get around to using it.
|
||||
// => https://github.com/neondatabase/neon/issues/6065
|
||||
slot_guard.file.replace(file);
|
||||
|
||||
Ok(vfile)
|
||||
}
|
||||
|
||||
/// Writes a file to the specified `final_path` in a crash safe fasion
|
||||
@@ -340,15 +339,15 @@ impl VirtualFile {
|
||||
));
|
||||
};
|
||||
std::fs::remove_file(tmp_path).or_else(fs_ext::ignore_not_found)?;
|
||||
let mut file = {
|
||||
let mut options = tokio_epoll_uring::ops::open_at::OpenOptions::new();
|
||||
options
|
||||
let mut file = Self::open_with_options(
|
||||
tmp_path,
|
||||
OpenOptions::new()
|
||||
.write(true)
|
||||
// Use `create_new` so that, if we race with ourselves or something else,
|
||||
// we bail out instead of causing damage.
|
||||
.create_new(true);
|
||||
Self::open_with_options_async(tmp_path, options).await?
|
||||
};
|
||||
.create_new(true),
|
||||
)
|
||||
.await?;
|
||||
file.write_all(content).await?;
|
||||
file.sync_all().await?;
|
||||
drop(file); // before the rename, that's important!
|
||||
@@ -359,93 +358,30 @@ impl VirtualFile {
|
||||
// the current `find_victim_slot` impl might pick the same slot for both
|
||||
// VirtualFile., and it eventually does a blocking write lock instead of
|
||||
// try_lock.
|
||||
let final_parent_dirfd = {
|
||||
let mut options = tokio_epoll_uring::ops::open_at::OpenOptions::new();
|
||||
options.read(true);
|
||||
Self::open_with_options_async(final_path_parent, options).await?
|
||||
};
|
||||
let final_parent_dirfd =
|
||||
Self::open_with_options(final_path_parent, OpenOptions::new().read(true)).await?;
|
||||
final_parent_dirfd.sync_all().await?;
|
||||
Ok(())
|
||||
}
|
||||
|
||||
/// Open a file with given options.
|
||||
///
|
||||
/// Note: If any custom flags were set in 'open_options' through OpenOptionsExt,
|
||||
/// they will be applied also when the file is subsequently re-opened, not only
|
||||
/// on the first time. Make sure that's sane!
|
||||
pub async fn open_with_options_async(
|
||||
path: &Utf8Path,
|
||||
open_options: tokio_epoll_uring::ops::open_at::OpenOptions,
|
||||
) -> Result<VirtualFile, std::io::Error> {
|
||||
let path_str = path.to_string();
|
||||
let parts = path_str.split('/').collect::<Vec<&str>>();
|
||||
let tenant_id;
|
||||
let timeline_id;
|
||||
if parts.len() > 5 && parts[parts.len() - 5] == "tenants" {
|
||||
tenant_id = parts[parts.len() - 4].to_string();
|
||||
timeline_id = parts[parts.len() - 2].to_string();
|
||||
} else {
|
||||
tenant_id = "*".to_string();
|
||||
timeline_id = "*".to_string();
|
||||
}
|
||||
let (handle, mut slot_guard) = get_open_files().find_victim_slot().await;
|
||||
|
||||
slot_guard.file = Some(observe_duration!(StorageIoOperation::Open, {
|
||||
let system = tokio_epoll_uring::thread_local_system().await;
|
||||
let file: OwnedFd = system
|
||||
.open(path, &open_options)
|
||||
.await
|
||||
.map_err(|e| match e {
|
||||
tokio_epoll_uring::Error::Op(e) => e,
|
||||
tokio_epoll_uring::Error::System(system) => {
|
||||
std::io::Error::new(std::io::ErrorKind::Other, system)
|
||||
}
|
||||
})?;
|
||||
File::from(file)
|
||||
}));
|
||||
|
||||
// Strip all options other than read and write.
|
||||
//
|
||||
// It would perhaps be nicer to check just for the read and write flags
|
||||
// explicitly, but OpenOptions doesn't contain any functions to read flags,
|
||||
// only to set them.
|
||||
let mut reopen_options = open_options;
|
||||
reopen_options.create(false);
|
||||
reopen_options.create_new(false);
|
||||
reopen_options.truncate(false);
|
||||
|
||||
let vfile = VirtualFile {
|
||||
handle: RwLock::new(handle),
|
||||
pos: 0,
|
||||
path: path.to_path_buf(),
|
||||
open_options: reopen_options,
|
||||
tenant_id,
|
||||
timeline_id,
|
||||
};
|
||||
|
||||
Ok(vfile)
|
||||
}
|
||||
|
||||
/// Call File::sync_all() on the underlying File.
|
||||
pub async fn sync_all(&self) -> Result<(), Error> {
|
||||
with_file!(self, StorageIoOperation::Fsync, |file| file
|
||||
.as_ref()
|
||||
.sync_all())
|
||||
self.with_file(StorageIoOperation::Fsync, |file| file.sync_all())
|
||||
.await?
|
||||
}
|
||||
|
||||
pub async fn metadata(&self) -> Result<fs::Metadata, Error> {
|
||||
with_file!(self, StorageIoOperation::Metadata, |file| file
|
||||
.as_ref()
|
||||
.metadata())
|
||||
self.with_file(StorageIoOperation::Metadata, |file| file.metadata())
|
||||
.await?
|
||||
}
|
||||
|
||||
/// Helper function internal to `VirtualFile` that looks up the underlying File,
|
||||
/// opens it and evicts some other File if necessary. The passed parameter is
|
||||
/// assumed to be a function available for the physical `File`.
|
||||
///
|
||||
/// We are doing it via a macro as Rust doesn't support async closures that
|
||||
/// take on parameters with lifetimes.
|
||||
async fn lock_file(&self) -> Result<FileGuard<'static>, Error> {
|
||||
/// Helper function that looks up the underlying File for this VirtualFile,
|
||||
/// opening it and evicting some other File if necessary. It calls 'func'
|
||||
/// with the physical File.
|
||||
async fn with_file<F, R>(&self, op: StorageIoOperation, mut func: F) -> Result<R, Error>
|
||||
where
|
||||
F: FnMut(&File) -> R,
|
||||
{
|
||||
let open_files = get_open_files();
|
||||
|
||||
let mut handle_guard = {
|
||||
@@ -455,23 +391,27 @@ impl VirtualFile {
|
||||
// We only need to hold the handle lock while we read the current handle. If
|
||||
// another thread closes the file and recycles the slot for a different file,
|
||||
// we will notice that the handle we read is no longer valid and retry.
|
||||
let mut handle = *self.handle.read().await;
|
||||
let mut handle = *self.handle.read().unwrap();
|
||||
loop {
|
||||
// Check if the slot contains our File
|
||||
{
|
||||
let slot = &open_files.slots[handle.index];
|
||||
let slot_guard = slot.inner.read().await;
|
||||
if slot_guard.tag == handle.tag && slot_guard.file.is_some() {
|
||||
// Found a cached file descriptor.
|
||||
slot.recently_used.store(true, Ordering::Relaxed);
|
||||
return Ok(FileGuard { slot_guard });
|
||||
let slot_guard = slot.inner.read().unwrap();
|
||||
if slot_guard.tag == handle.tag {
|
||||
if let Some(file) = &slot_guard.file {
|
||||
// Found a cached file descriptor.
|
||||
slot.recently_used.store(true, Ordering::Relaxed);
|
||||
return Ok(STORAGE_IO_TIME_METRIC
|
||||
.get(op)
|
||||
.observe_closure_duration(|| func(file)));
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// The slot didn't contain our File. We will have to open it ourselves,
|
||||
// but before that, grab a write lock on handle in the VirtualFile, so
|
||||
// that no other thread will try to concurrently open the same file.
|
||||
let handle_guard = self.handle.write().await;
|
||||
let handle_guard = self.handle.write().unwrap();
|
||||
|
||||
// If another thread changed the handle while we were not holding the lock,
|
||||
// then the handle might now be valid again. Loop back to retry.
|
||||
@@ -485,25 +425,20 @@ impl VirtualFile {
|
||||
|
||||
// We need to open the file ourselves. The handle in the VirtualFile is
|
||||
// now locked in write-mode. Find a free slot to put it in.
|
||||
let (handle, mut slot_guard) = open_files.find_victim_slot().await;
|
||||
let (handle, mut slot_guard) = open_files.find_victim_slot();
|
||||
|
||||
// Re-open the physical file.
|
||||
// NB: we use StorageIoOperation::OpenAferReplace for this to distinguish this
|
||||
// case from StorageIoOperation::Open. This helps with identifying thrashing
|
||||
// of the virtual file descriptor cache.
|
||||
let file = observe_duration!(StorageIoOperation::OpenAfterReplace, {
|
||||
let system = tokio_epoll_uring::thread_local_system().await;
|
||||
let file: OwnedFd = system
|
||||
.open(&self.path, &self.open_options)
|
||||
.await
|
||||
.map_err(|e| match e {
|
||||
tokio_epoll_uring::Error::Op(e) => e,
|
||||
tokio_epoll_uring::Error::System(system) => {
|
||||
std::io::Error::new(std::io::ErrorKind::Other, system)
|
||||
}
|
||||
})?;
|
||||
File::from(file)
|
||||
});
|
||||
let file = STORAGE_IO_TIME_METRIC
|
||||
.get(StorageIoOperation::OpenAfterReplace)
|
||||
.observe_closure_duration(|| self.open_options.open(&self.path))?;
|
||||
|
||||
// Perform the requested operation on it
|
||||
let result = STORAGE_IO_TIME_METRIC
|
||||
.get(op)
|
||||
.observe_closure_duration(|| func(&file));
|
||||
|
||||
// Store the File in the slot and update the handle in the VirtualFile
|
||||
// to point to it.
|
||||
@@ -511,9 +446,7 @@ impl VirtualFile {
|
||||
|
||||
*handle_guard = handle;
|
||||
|
||||
return Ok(FileGuard {
|
||||
slot_guard: slot_guard.downgrade(),
|
||||
});
|
||||
Ok(result)
|
||||
}
|
||||
|
||||
pub fn remove(self) {
|
||||
@@ -528,9 +461,11 @@ impl VirtualFile {
|
||||
self.pos = offset;
|
||||
}
|
||||
SeekFrom::End(offset) => {
|
||||
self.pos = with_file!(self, StorageIoOperation::Seek, |file| file
|
||||
.as_ref()
|
||||
.seek(SeekFrom::End(offset)))?
|
||||
self.pos = self
|
||||
.with_file(StorageIoOperation::Seek, |mut file| {
|
||||
file.seek(SeekFrom::End(offset))
|
||||
})
|
||||
.await??
|
||||
}
|
||||
SeekFrom::Current(offset) => {
|
||||
let pos = self.pos as i128 + offset as i128;
|
||||
@@ -550,48 +485,24 @@ impl VirtualFile {
|
||||
}
|
||||
|
||||
// Copied from https://doc.rust-lang.org/1.72.0/src/std/os/unix/fs.rs.html#117-135
|
||||
pub async fn read_exact_at<B>(&self, buf: B, mut offset: u64) -> Result<B, Error>
|
||||
where
|
||||
B: IoBufMut + Send,
|
||||
{
|
||||
use tokio_epoll_uring::BoundedBuf;
|
||||
let mut buf: tokio_epoll_uring::Slice<B> = buf.slice_full();
|
||||
while buf.bytes_total() != 0 {
|
||||
let res;
|
||||
(buf, res) = self.read_at(buf, offset).await;
|
||||
match res {
|
||||
Ok(0) => break,
|
||||
pub async fn read_exact_at(&self, mut buf: &mut [u8], mut offset: u64) -> Result<(), Error> {
|
||||
while !buf.is_empty() {
|
||||
match self.read_at(buf, offset).await {
|
||||
Ok(0) => {
|
||||
return Err(Error::new(
|
||||
std::io::ErrorKind::UnexpectedEof,
|
||||
"failed to fill whole buffer",
|
||||
))
|
||||
}
|
||||
Ok(n) => {
|
||||
buf = buf.slice(n..);
|
||||
buf = &mut buf[n..];
|
||||
offset += n as u64;
|
||||
}
|
||||
Err(ref e) if e.kind() == std::io::ErrorKind::Interrupted => {}
|
||||
Err(e) => return Err(e),
|
||||
}
|
||||
}
|
||||
if !buf.is_empty() {
|
||||
Err(std::io::Error::new(
|
||||
std::io::ErrorKind::UnexpectedEof,
|
||||
"failed to fill whole buffer",
|
||||
))
|
||||
} else {
|
||||
Ok(buf.into_inner())
|
||||
}
|
||||
}
|
||||
|
||||
/// Like [`Self::read_exact_at`] but for [`PageWriteGuard`].
|
||||
pub async fn read_exact_at_page(
|
||||
&self,
|
||||
page: PageWriteGuard<'static>,
|
||||
offset: u64,
|
||||
) -> Result<PageWriteGuard<'static>, Error> {
|
||||
let buf = PageWriteGuardBuf {
|
||||
page,
|
||||
init_up_to: 0,
|
||||
};
|
||||
let res = self.read_exact_at(buf, offset).await;
|
||||
res.map(|PageWriteGuardBuf { page, .. }| page)
|
||||
.map_err(|e| Error::new(ErrorKind::Other, e))
|
||||
Ok(())
|
||||
}
|
||||
|
||||
// Copied from https://doc.rust-lang.org/1.72.0/src/std/os/unix/fs.rs.html#219-235
|
||||
@@ -641,59 +552,23 @@ impl VirtualFile {
|
||||
Ok(n)
|
||||
}
|
||||
|
||||
pub(crate) async fn read_at<B>(&self, buf: B, offset: u64) -> (B, Result<usize, Error>)
|
||||
where
|
||||
B: tokio_epoll_uring::BoundedBufMut + Send,
|
||||
{
|
||||
let file_guard = match self.lock_file().await {
|
||||
Ok(file_guard) => file_guard,
|
||||
Err(e) => return (buf, Err(e)),
|
||||
};
|
||||
let (buf, result) = observe_duration!(StorageIoOperation::Read, {
|
||||
self.read_at0(file_guard, buf, offset).await
|
||||
});
|
||||
(buf, result)
|
||||
}
|
||||
async fn read_at0<B>(
|
||||
&self,
|
||||
file_guard: FileGuard<'_>,
|
||||
buf: B,
|
||||
offset: u64,
|
||||
) -> (B, Result<usize, Error>)
|
||||
where
|
||||
B: tokio_epoll_uring::BoundedBufMut + Send,
|
||||
{
|
||||
let system = tokio_epoll_uring::thread_local_system().await;
|
||||
// SAFETY: when file_guard gets dropped, the raw fd becomes invalid or may get re-used
|
||||
// while the io_uring operation is still executing.
|
||||
// The `file_guard` could get dropped due to future cancellation-by-drop.
|
||||
// We prevent this situation using the scopeguard: it will abort the process in such cases.
|
||||
// Fixing this is subject of https://github.com/neondatabase/neon/pull/6101
|
||||
let owned_fd = unsafe { OwnedFd::from_raw_fd(file_guard.as_ref().as_raw_fd()) };
|
||||
let guard = scopeguard::guard(file_guard, |_| {
|
||||
eprintln!("must not drop future while operation is ongoing (todo: pass file_guard to tokio_epoll_uring to aovid this)");
|
||||
std::process::abort();
|
||||
});
|
||||
let ((owned_fd, buf), res) = system.read(owned_fd, offset, buf).await;
|
||||
let _ = OwnedFd::into_raw_fd(owned_fd);
|
||||
let _ = scopeguard::ScopeGuard::into_inner(guard);
|
||||
if let Ok(size) = res {
|
||||
// TODO: don't use with_label_values on hot path
|
||||
// https://github.com/neondatabase/neon/issues/6107
|
||||
pub async fn read_at(&self, buf: &mut [u8], offset: u64) -> Result<usize, Error> {
|
||||
let result = self
|
||||
.with_file(StorageIoOperation::Read, |file| file.read_at(buf, offset))
|
||||
.await?;
|
||||
if let Ok(size) = result {
|
||||
STORAGE_IO_SIZE
|
||||
.with_label_values(&["read", &self.tenant_id, &self.timeline_id])
|
||||
.add(size as i64);
|
||||
}
|
||||
(buf, res.map_err(|e| Error::new(ErrorKind::Other, e)))
|
||||
result
|
||||
}
|
||||
|
||||
async fn write_at(&self, buf: &[u8], offset: u64) -> Result<usize, Error> {
|
||||
let result = with_file!(self, StorageIoOperation::Write, |file| file
|
||||
.as_ref()
|
||||
.write_at(buf, offset));
|
||||
let result = self
|
||||
.with_file(StorageIoOperation::Write, |file| file.write_at(buf, offset))
|
||||
.await?;
|
||||
if let Ok(size) = result {
|
||||
// TODO: don't use with_label_values on hot path
|
||||
// https://github.com/neondatabase/neon/issues/6107
|
||||
STORAGE_IO_SIZE
|
||||
.with_label_values(&["write", &self.tenant_id, &self.timeline_id])
|
||||
.add(size as i64);
|
||||
@@ -702,18 +577,6 @@ impl VirtualFile {
|
||||
}
|
||||
}
|
||||
|
||||
struct FileGuard<'a> {
|
||||
slot_guard: RwLockReadGuard<'a, SlotInner>,
|
||||
}
|
||||
|
||||
impl<'a> AsRef<File> for FileGuard<'a> {
|
||||
fn as_ref(&self) -> &File {
|
||||
// This unwrap is safe because we only create `FileGuard`s
|
||||
// if we know that the file is Some.
|
||||
self.slot_guard.file.as_ref().unwrap()
|
||||
}
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
impl VirtualFile {
|
||||
pub(crate) async fn read_blk(
|
||||
@@ -721,19 +584,16 @@ impl VirtualFile {
|
||||
blknum: u32,
|
||||
) -> Result<crate::tenant::block_io::BlockLease<'_>, std::io::Error> {
|
||||
use crate::page_cache::PAGE_SZ;
|
||||
let buf = vec![0; PAGE_SZ];
|
||||
let buf = self
|
||||
.read_exact_at(buf, blknum as u64 * (PAGE_SZ as u64))
|
||||
let mut buf = [0; PAGE_SZ];
|
||||
self.read_exact_at(&mut buf, blknum as u64 * (PAGE_SZ as u64))
|
||||
.await?;
|
||||
Ok(crate::tenant::block_io::BlockLease::Vec(buf))
|
||||
Ok(std::sync::Arc::new(buf).into())
|
||||
}
|
||||
|
||||
async fn read_to_end(&mut self, buf: &mut Vec<u8>) -> Result<(), Error> {
|
||||
let mut tmp = vec![0; 128];
|
||||
loop {
|
||||
let res;
|
||||
(tmp, res) = self.read_at(tmp, self.pos).await;
|
||||
match res {
|
||||
let mut tmp = [0; 128];
|
||||
match self.read_at(&mut tmp, self.pos).await {
|
||||
Ok(0) => return Ok(()),
|
||||
Ok(n) => {
|
||||
self.pos += n as u64;
|
||||
@@ -749,41 +609,22 @@ impl VirtualFile {
|
||||
impl Drop for VirtualFile {
|
||||
/// If a VirtualFile is dropped, close the underlying file if it was open.
|
||||
fn drop(&mut self) {
|
||||
let handle = self.handle.get_mut();
|
||||
let handle = self.handle.get_mut().unwrap();
|
||||
|
||||
fn clean_slot(slot: &Slot, mut slot_guard: RwLockWriteGuard<'_, SlotInner>, tag: u64) {
|
||||
if slot_guard.tag == tag {
|
||||
slot.recently_used.store(false, Ordering::Relaxed);
|
||||
// there is also the `CloseByReplace` operation for closes done on eviction for
|
||||
// comparison.
|
||||
if let Some(fd) = slot_guard.file.take() {
|
||||
STORAGE_IO_TIME_METRIC
|
||||
.get(StorageIoOperation::Close)
|
||||
.observe_closure_duration(|| drop(fd));
|
||||
}
|
||||
// We could check with a read-lock first, to avoid waiting on an
|
||||
// unrelated I/O.
|
||||
let slot = &get_open_files().slots[handle.index];
|
||||
let mut slot_guard = slot.inner.write().unwrap();
|
||||
if slot_guard.tag == handle.tag {
|
||||
slot.recently_used.store(false, Ordering::Relaxed);
|
||||
// there is also operation "close-by-replace" for closes done on eviction for
|
||||
// comparison.
|
||||
if let Some(fd) = slot_guard.file.take() {
|
||||
STORAGE_IO_TIME_METRIC
|
||||
.get(StorageIoOperation::Close)
|
||||
.observe_closure_duration(|| drop(fd));
|
||||
}
|
||||
}
|
||||
|
||||
// We don't have async drop so we cannot directly await the lock here.
|
||||
// Instead, first do a best-effort attempt at closing the underlying
|
||||
// file descriptor by using `try_write`, and if that fails, spawn
|
||||
// a tokio task to do it asynchronously: we just want it to be
|
||||
// cleaned up eventually.
|
||||
// Most of the time, the `try_lock` should succeed though,
|
||||
// as we have `&mut self` access. In other words, if the slot
|
||||
// is still occupied by our file, there should be no access from
|
||||
// other I/O operations; the only other possible place to lock
|
||||
// the slot is the lock algorithm looking for free slots.
|
||||
let slot = &get_open_files().slots[handle.index];
|
||||
if let Ok(slot_guard) = slot.inner.try_write() {
|
||||
clean_slot(slot, slot_guard, handle.tag);
|
||||
} else {
|
||||
let tag = handle.tag;
|
||||
tokio::spawn(async move {
|
||||
let slot_guard = slot.inner.write().await;
|
||||
clean_slot(slot, slot_guard, tag);
|
||||
});
|
||||
};
|
||||
}
|
||||
}
|
||||
|
||||
@@ -857,10 +698,10 @@ mod tests {
|
||||
}
|
||||
|
||||
impl MaybeVirtualFile {
|
||||
async fn read_exact_at(&self, mut buf: Vec<u8>, offset: u64) -> Result<Vec<u8>, Error> {
|
||||
async fn read_exact_at(&self, buf: &mut [u8], offset: u64) -> Result<(), Error> {
|
||||
match self {
|
||||
MaybeVirtualFile::VirtualFile(file) => file.read_exact_at(buf, offset).await,
|
||||
MaybeVirtualFile::File(file) => file.read_exact_at(&mut buf, offset).map(|()| buf),
|
||||
MaybeVirtualFile::File(file) => file.read_exact_at(buf, offset),
|
||||
}
|
||||
}
|
||||
async fn write_all_at(&self, buf: &[u8], offset: u64) -> Result<(), Error> {
|
||||
@@ -902,15 +743,14 @@ mod tests {
|
||||
|
||||
// Helper function to slurp a portion of a file into a string
|
||||
async fn read_string_at(&mut self, pos: u64, len: usize) -> Result<String, Error> {
|
||||
let buf = vec![0; len];
|
||||
let buf = self.read_exact_at(buf, pos).await?;
|
||||
let mut buf = vec![0; len];
|
||||
self.read_exact_at(&mut buf, pos).await?;
|
||||
Ok(String::from_utf8(buf).unwrap())
|
||||
}
|
||||
}
|
||||
|
||||
#[tokio::test]
|
||||
async fn test_virtual_files() -> anyhow::Result<()> {
|
||||
crate::tenant::harness::setup_logging();
|
||||
async fn test_virtual_files() -> Result<(), Error> {
|
||||
// The real work is done in the test_files() helper function. This
|
||||
// allows us to run the same set of tests against a native File, and
|
||||
// VirtualFile. We trust the native Files and wouldn't need to test them,
|
||||
@@ -919,35 +759,23 @@ mod tests {
|
||||
// native files, you will run out of file descriptors if the ulimit
|
||||
// is low enough.)
|
||||
test_files("virtual_files", |path, open_options| async move {
|
||||
let vf = VirtualFile::open_with_options_async(&path, open_options).await?;
|
||||
let vf = VirtualFile::open_with_options(&path, &open_options).await?;
|
||||
Ok(MaybeVirtualFile::VirtualFile(vf))
|
||||
})
|
||||
.await
|
||||
}
|
||||
|
||||
#[tokio::test]
|
||||
async fn test_physical_files() -> anyhow::Result<()> {
|
||||
async fn test_physical_files() -> Result<(), Error> {
|
||||
test_files("physical_files", |path, open_options| async move {
|
||||
Ok(MaybeVirtualFile::File({
|
||||
let system = tokio_epoll_uring::thread_local_system().await;
|
||||
let owned_fd = system
|
||||
.open(path, &open_options)
|
||||
.await
|
||||
.map_err(|e| match e {
|
||||
tokio_epoll_uring::Error::Op(e) => e,
|
||||
tokio_epoll_uring::Error::System(system) => {
|
||||
std::io::Error::new(std::io::ErrorKind::Other, system)
|
||||
}
|
||||
})?;
|
||||
File::from(owned_fd)
|
||||
}))
|
||||
Ok(MaybeVirtualFile::File(open_options.open(path)?))
|
||||
})
|
||||
.await
|
||||
}
|
||||
|
||||
async fn test_files<OF, FT>(testname: &str, openfunc: OF) -> anyhow::Result<()>
|
||||
async fn test_files<OF, FT>(testname: &str, openfunc: OF) -> Result<(), Error>
|
||||
where
|
||||
OF: Fn(Utf8PathBuf, tokio_epoll_uring::ops::open_at::OpenOptions) -> FT,
|
||||
OF: Fn(Utf8PathBuf, OpenOptions) -> FT,
|
||||
FT: Future<Output = Result<MaybeVirtualFile, std::io::Error>>,
|
||||
{
|
||||
let testdir = crate::config::PageServerConf::test_repo_dir(testname);
|
||||
@@ -956,7 +784,7 @@ mod tests {
|
||||
let path_a = testdir.join("file_a");
|
||||
let mut file_a = openfunc(
|
||||
path_a.clone(),
|
||||
tokio_epoll_uring::ops::open_at::OpenOptions::new()
|
||||
OpenOptions::new()
|
||||
.write(true)
|
||||
.create(true)
|
||||
.truncate(true)
|
||||
@@ -969,13 +797,7 @@ mod tests {
|
||||
let _ = file_a.read_string().await.unwrap_err();
|
||||
|
||||
// Close the file and re-open for reading
|
||||
let mut file_a = openfunc(
|
||||
path_a,
|
||||
tokio_epoll_uring::ops::open_at::OpenOptions::new()
|
||||
.read(true)
|
||||
.to_owned(),
|
||||
)
|
||||
.await?;
|
||||
let mut file_a = openfunc(path_a, OpenOptions::new().read(true).to_owned()).await?;
|
||||
|
||||
// cannot write to a file opened in read-only mode
|
||||
let _ = file_a.write_all(b"bar").await.unwrap_err();
|
||||
@@ -1012,7 +834,7 @@ mod tests {
|
||||
let path_b = testdir.join("file_b");
|
||||
let mut file_b = openfunc(
|
||||
path_b.clone(),
|
||||
tokio_epoll_uring::ops::open_at::OpenOptions::new()
|
||||
OpenOptions::new()
|
||||
.read(true)
|
||||
.write(true)
|
||||
.create(true)
|
||||
@@ -1033,13 +855,8 @@ mod tests {
|
||||
|
||||
let mut vfiles = Vec::new();
|
||||
for _ in 0..100 {
|
||||
let mut vfile = openfunc(
|
||||
path_b.clone(),
|
||||
tokio_epoll_uring::ops::open_at::OpenOptions::new()
|
||||
.read(true)
|
||||
.to_owned(),
|
||||
)
|
||||
.await?;
|
||||
let mut vfile =
|
||||
openfunc(path_b.clone(), OpenOptions::new().read(true).to_owned()).await?;
|
||||
assert_eq!("FOOBAR", vfile.read_string().await?);
|
||||
vfiles.push(vfile);
|
||||
}
|
||||
@@ -1084,12 +901,8 @@ mod tests {
|
||||
// Open the file many times.
|
||||
let mut files = Vec::new();
|
||||
for _ in 0..VIRTUAL_FILES {
|
||||
let f = VirtualFile::open_with_options_async(&test_file_path, {
|
||||
let mut options = tokio_epoll_uring::ops::open_at::OpenOptions::new();
|
||||
options.read(true);
|
||||
options
|
||||
})
|
||||
.await?;
|
||||
let f = VirtualFile::open_with_options(&test_file_path, OpenOptions::new().read(true))
|
||||
.await?;
|
||||
files.push(f);
|
||||
}
|
||||
let files = Arc::new(files);
|
||||
@@ -1104,11 +917,11 @@ mod tests {
|
||||
for _threadno in 0..THREADS {
|
||||
let files = files.clone();
|
||||
let hdl = rt.spawn(async move {
|
||||
let mut buf = vec![0u8; SIZE];
|
||||
let mut buf = [0u8; SIZE];
|
||||
let mut rng = rand::rngs::OsRng;
|
||||
for _ in 1..1000 {
|
||||
let f = &files[rng.gen_range(0..files.len())];
|
||||
buf = f.read_exact_at(buf, 0).await.unwrap();
|
||||
f.read_exact_at(&mut buf, 0).await.unwrap();
|
||||
assert!(buf == SAMPLE);
|
||||
}
|
||||
});
|
||||
|
||||
Reference in New Issue
Block a user