-
-
Notifications
You must be signed in to change notification settings - Fork 313
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Smart-GC - build a big pack from smaller ones (incrementally) as file handles get scarce #15
Comments
I think that would be the ideal solution for docs.rs. Right now we have giant spikes in the number of files opened every time we update the index: It's gone down a lot since we started running GC (it used to be 3k files!) but it's still fairly high. I didn't realize crates-index-diff was loading every file, that would explain it :P There might still need to be some way to GC since otherwise the number of packfiles would keep going up over time and operations on the index would get slower.
That wouldn't be great, since (for docs.rs) it means that we can't serve any web requests until gitoxide notices the difference and finishes performing the GC - we only call it every minute. |
Thanks for chiming in! I particularly like to see these spikes for once with my own eyes! Curious as to how that will look like with
No no :), it doesn't. All it does it use the high-level git2 API to fetch and to diff.
True, smart really needs to be that, and to 'the right thing™️' to not loose performance noticeably over time. I yet have to see how performance turns out for common operations as well.
In my perfect naive world, it can do an incremental GC to avoid slowdowns and serve the request in the same or less time than libgit2, so it appears entirely free. |
❤️
Oh I was confusing myself - we'd only hit the file limit when we call crates-index-diff. So if it can notice hitting the limit automatically then it can run gc immediately, and there'd be very little time when we were at the limit. I'd still prefer for it to be no time, but that sounds like it could work. |
I think I will (in time) be writing a test for this - clone a the index locally at a long past commit, and simulate the parent repository gaining a few commits, then pull, and repeat while watching how files pile up in the current implementation. Repeat with |
The new URL should trigger an overflow check but it only happens when `url::Url::parse()` is called directly as our code doesn't let it through anymore. Here is the log from the fuzzer run as reported: ``` [Environment] ASAN_OPTIONS=handle_abort=2 +----------------------------------------Release Build Stacktrace----------------------------------------+ Command: /mnt/scratch0/clusterfuzz/resources/platform/linux/unshare -c -n /mnt/scratch0/clusterfuzz/bot/builds/clusterfuzz-builds_gitoxide_9a561c2a19701ceb3cded247e9ae8f349711bbca/revisions/gix-url-parse -rss_limit_mb=2560 -timeout=60 -runs=100 /mnt/scratch0/clusterfuzz/bot/inputs/fuzzer-testcases/f508abd59698de9914f2b8894cc135f55208e494873d456c6c19828509103805 Time ran: 0.12717413902282715 INFO: Running with entropic power schedule (0xFF, 100). INFO: Seed: 1001182178 INFO: Loaded 1 modules (90683 inline 8-bit counters): 90683 [0x5d0c0597cce0, 0x5d0c05992f1b), INFO: Loaded 1 PC tables (90683 PCs): 90683 [0x5d0c05992f20,0x5d0c05af52d0), /mnt/scratch0/clusterfuzz/bot/builds/clusterfuzz-builds_gitoxide_9a561c2a19701ceb3cded247e9ae8f349711bbca/revisions/gix-url-parse: Running 1 inputs 100 time(s) each. Running: /mnt/scratch0/clusterfuzz/bot/inputs/fuzzer-testcases/f508abd59698de9914f2b8894cc135f55208e494873d456c6c19828509103805 thread '<unnamed>' panicked at /rust/registry/src/index.crates.io-6f17d22bba15001f/idna-0.4.0/src/punycode.rs:272:17: attempt to add with overflow note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace AddressSanitizer:DEADLYSIGNAL ================================================================= ==1200==ERROR: AddressSanitizer: ABRT on unknown address 0x0539000004b0 (pc 0x7c51742fb00b bp 0x7ffcd1eadd80 sp 0x7ffcd1eadaf0 T0) #0 0x7c51742fb00b in raise /build/glibc-SzIz7B/glibc-2.31/sysdeps/unix/sysv/linux/raise.c:51:1 #1 0x7c51742da858 in abort /build/glibc-SzIz7B/glibc-2.31/stdlib/abort.c:79:7 #2 0x5d0c0572a1e6 in std::sys::unix::abort_internal::he854d2f74b119e66 /rustc/e20cb7702117f1ad8127a16406ba9edd230c4f65/library/std/src/sys/unix/mod.rs:375:14 #3 0x5d0c0518cda6 in std::process::abort::h68c27a968dc7c74f /rustc/e20cb7702117f1ad8127a16406ba9edd230c4f65/library/std/src/process.rs:2271:5 #4 0x5d0c0564d3d4 in libfuzzer_sys::initialize::_$u7b$$u7b$closure$u7d$$u7d$::h1e76e422e0c48db0 /rust/registry/src/index.crates.io-6f17d22bba15001f/libfuzzer-sys-0.4.3/src/lib.rs:57:9 #5 0x5d0c0571dcf7 in _$LT$alloc..boxed..Box$LT$F$C$A$GT$$u20$as$u20$core..ops..function..Fn$LT$Args$GT$$GT$::call::h0c028c5af3475e03 /rustc/e20cb7702117f1ad8127a16406ba9edd230c4f65/library/alloc/src/boxed.rs:2021:9 #6 0x5d0c0571dcf7 in std::panicking::rust_panic_with_hook::hd26c5407fbf20d71 /rustc/e20cb7702117f1ad8127a16406ba9edd230c4f65/library/std/src/panicking.rs:735:13 #7 0x5d0c0571da05 in std::panicking::begin_panic_handler::_$u7b$$u7b$closure$u7d$$u7d$::h944e23ea90982f5a /rustc/e20cb7702117f1ad8127a16406ba9edd230c4f65/library/std/src/panicking.rs:601:13 #8 0x5d0c0571aee5 in std::sys_common::backtrace::__rust_end_short_backtrace::h8a3632d339dd3313 /rustc/e20cb7702117f1ad8127a16406ba9edd230c4f65/library/std/src/sys_common/backtrace.rs:170:18 #9 0x5d0c0571d781 in rust_begin_unwind /rustc/e20cb7702117f1ad8127a16406ba9edd230c4f65/library/std/src/panicking.rs:597:5 #10 0x5d0c05190634 in core::panicking::panic_fmt::h85c36fc727234039 /rustc/e20cb7702117f1ad8127a16406ba9edd230c4f65/library/core/src/panicking.rs:72:14 #11 0x5d0c051906d2 in core::panicking::panic::h6a47ed7881a36f4d /rustc/e20cb7702117f1ad8127a16406ba9edd230c4f65/library/core/src/panicking.rs:127:5 #12 0x5d0c053e09c6 in idna::punycode::encode_into::hd674630fb161bf5b /rust/registry/src/index.crates.io-6f17d22bba15001f/idna-0.4.0/src/punycode.rs:0 #13 0x5d0c053eacbc in idna::uts46::Idna::to_ascii_inner::h69c52eb69ae48276 /rust/registry/src/index.crates.io-6f17d22bba15001f/idna-0.4.0/src/uts46.rs:469:34 #14 0x5d0c053eb793 in idna::uts46::Idna::to_ascii::h76237795045112f3 /rust/registry/src/index.crates.io-6f17d22bba15001f/idna-0.4.0/src/uts46.rs:481:26 #15 0x5d0c053eda7a in idna::uts46::Config::to_ascii::h423c722ab2fa9813 /rust/registry/src/index.crates.io-6f17d22bba15001f/idna-0.4.0/src/uts46.rs:572:9 #16 0x5d0c053f0070 in idna::domain_to_ascii::h93e94e995d03e9ef /rust/registry/src/index.crates.io-6f17d22bba15001f/idna-0.4.0/src/lib.rs:64:5 #17 0x5d0c0530374a in url::host::Host::domain_to_ascii::h6cb1ae8fe42a1e42 /rust/registry/src/index.crates.io-6f17d22bba15001f/url-2.4.1/src/host.rs:166:9 #18 0x5d0c0530374a in url::host::Host::parse::h962d3990e0ff5091 /rust/registry/src/index.crates.io-6f17d22bba15001f/url-2.4.1/src/host.rs:86:22 #19 0x5d0c0532e87d in url::parser::Parser::parse_host::h89faea9182ce2512 /rust/registry/src/index.crates.io-6f17d22bba15001f/url-2.4.1/src/parser.rs:1024:20 #20 0x5d0c0532ba8d in url::parser::Parser::parse_host_and_port::heb44bd7ebd2593f6 /rust/registry/src/index.crates.io-6f17d22bba15001f/url-2.4.1/src/parser.rs:944:33 #21 0x5d0c0532896f in url::parser::Parser::after_double_slash::hbb313f562f0978a2 /rust/registry/src/index.crates.io-6f17d22bba15001f/url-2.4.1/src/parser.rs:843:13 #22 0x5d0c0531a129 in url::parser::Parser::parse_with_scheme::h54a417e4650ea024 /rust/registry/src/index.crates.io-6f17d22bba15001f/url-2.4.1/src/parser.rs:453:17 #23 0x5d0c05317824 in url::parser::Parser::parse_url::hfa6b21c53cd0ac1c /rust/registry/src/index.crates.io-6f17d22bba15001f/url-2.4.1/src/parser.rs:366:20 #24 0x5d0c05350ba0 in url::ParseOptions::parse::hb8b3309b3b920457 /rust/registry/src/index.crates.io-6f17d22bba15001f/url-2.4.1/src/lib.rs:257:9 #25 0x5d0c052b1ffc in url::Url::parse::h82a965c69df59bba /rust/registry/src/index.crates.io-6f17d22bba15001f/url-2.4.1/src/lib.rs:292:9 #26 0x5d0c052b1ffc in gix_url::parse::input_to_utf8_and_url::h14b70a32a8884316 gitoxide/gix-url/src/parse.rs:252:5 #27 0x5d0c052a9b9d in gix_url::parse::url::h6f7f7b0bddf4b8d7 gitoxide/gix-url/src/parse.rs:99:24 #28 0x5d0c052b34cc in gix_url::parse::hfaab74909f01c9cc gitoxide/gix-url/src/lib.rs:38:46 #29 0x5d0c05270b4a in rust_fuzzer_test_input gitoxide/gix-url/fuzz/fuzz_targets/parse.rs:5:14 #30 0x5d0c0564d537 in __rust_try libfuzzer_sys.f28e88650cadb2d4-cgu.0:0 #31 0x5d0c0564c79f in std::panicking::try::h90783eeef7e35925 /rustc/e20cb7702117f1ad8127a16406ba9edd230c4f65/library/std/src/panicking.rs:468:19 #32 0x5d0c0564c79f in std::panic::catch_unwind::h041b281a0e92d580 /rustc/e20cb7702117f1ad8127a16406ba9edd230c4f65/library/std/src/panic.rs:142:14 #33 0x5d0c0564c79f in LLVMFuzzerTestOneInput /rust/registry/src/index.crates.io-6f17d22bba15001f/libfuzzer-sys-0.4.3/src/lib.rs:28:22 #34 0x5d0c0566bb83 in fuzzer::Fuzzer::ExecuteCallback(unsigned char const*, unsigned long) /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerLoop.cpp:611:15 #35 0x5d0c056572e2 in fuzzer::RunOneTest(fuzzer::Fuzzer*, char const*, unsigned long) /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerDriver.cpp:324:6 #36 0x5d0c0565cb8c in fuzzer::FuzzerDriver(int*, char***, int (*)(unsigned char const*, unsigned long)) /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerDriver.cpp:860:9 #37 0x5d0c056860c2 in main /src/llvm-project/compiler-rt/lib/fuzzer/FuzzerMain.cpp:20:10 #38 0x7c51742dc082 in __libc_start_main /build/glibc-SzIz7B/glibc-2.31/csu/libc-start.c:308:16 #39 0x5d0c05191c4d in _start ``` `
It's too easy to get bitten by
git pull
not working anymore as git cannot perform its operations anymore. This is because each fetch MAY create a new pack in libgit2. And even if not, and packs are exploded into loose objects, usually 10_000 of these warrant the creation of a new pack. It's just a matter of time until there are too many of them for the git-repository to memory map all of these for fast access.Gitoxide should be better and either GC automatically once files cannot be opened anymore due to insufficient handles, or track its own usage to know when they are about to be tight and do an incremental GC to combine a few packs into one, fast.
Another avenue would be try to map only packs that actually have objects we are interested in, avoiding to map all by default, if that's even possible or viable.
Related to rust-lang/docs.rs#975 , and probably many more.
The text was updated successfully, but these errors were encountered: