-
Notifications
You must be signed in to change notification settings - Fork 589
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
Shard aware connections v2 #8
Shard aware connections v2 #8
Conversation
Seastar support different load balancing algorithms that determine which shard will handle incoming connection. Added configuration option to rpc::server allowing to configure the listener load balancing algorithm. Signed-off-by: Michal Maslanka <[email protected]>
Added current node_id shift to outgoing connections shard assignment. In order to make sure that connections to the same node created on different nodes are assigned to different shards added node_id based offset to connections shard assignment. This approach is required to make use of deterministic incoming connection to shard assignment on the target node. If there would be no shift all connections to given node would be handled on the same set of shards. Signed-off-by: Michal Maslanka <[email protected]>
Using port based shard connection assignment in RPC server. This makes incoming connection shard assignment deterministic. Signed-off-by: Michal Maslanka <[email protected]>
@@ -92,6 +99,8 @@ inline ss::shard_id connection_cache::shard_for( | |||
// NOLINTNEXTLINE | |||
size_t h = universe[jump_consistent_hash(src_shard, vnodes)]; | |||
boost::hash_combine(h, std::hash<model::node_id>{}(n)); | |||
boost::hash_combine(h, std::hash<model::node_id>{}(self)); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
THe only question I have here, is whether or not this actually makes a difference in the node balancing.
We have 3 sources of entropy
- total shards
- source shard
- destination node_id
and this adds the
- source node_id
What does the distribution look like
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Here is an example for 3 nodes cluster 16 core each:
Here number represent shards:
0 -> 1 = outgoing: [{1, 12, 13}] incoming: [{1, 12, 13}]
0 -> 2 = outgoing: [{0, 15}] incoming: [{0, 15}]
1 -> 0 = outgoing: [{8, 10, 3}] incoming: [{8, 10, 3}]
1 -> 2 = outgoing: [{0, 8, 15}] incoming: [{0, 8, 15}]
2 -> 0 = outgoing: [{10, 5}] incoming: [{10, 5}]
2 -> 1 = outgoing: [{15, 6, 7}] incoming: [{15, 6, 7}]
This is connections count per shard:
Node 0 clients shards {1: 1, 12: 1, 13: 1, 0: 1, 15: 1}
Node 1 clients shards {8: 2, 10: 1, 3: 1, 0: 1, 15: 1}
Node 2 clients shards {10: 1, 5: 1, 15: 1, 6: 1, 7: 1}
Node 1 server shards {1: 1, 12: 1, 13: 1, 15: 1, 6: 1, 7: 1}
Node 2 server shards {0: 2, 15: 2, 8: 1}
Node 0 server shards {8: 1, 10: 2, 3: 1, 5: 1}
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I don't think i understand the output.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
what is 2-1? or 0->1 ?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I guess i'm looking for a before and after
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
0 -> 1
are connections from node 0 to node 1, outgoing are shards on node 0, incoming are shards on node 1
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Oh I didn't notice that you are interested in change that is made
Assignment before:
0 -> 1 = outgoing: [{11, 15, 7}] incoming: [{11, 15, 7}]
0 -> 2 = outgoing: [{1, 10, 5}] incoming: [{1, 10, 5}]
1 -> 0 = outgoing: [{3, 5}] incoming: [{3, 5}]
1 -> 2 = outgoing: [{1, 10, 5}] incoming: [{1, 10, 5}]
2 -> 0 = outgoing: [{3, 5}] incoming: [{3, 5}]
2 -> 1 = outgoing: [{11, 15, 7}] incoming: [{11, 15, 7}]
After:
0 -> 1 = outgoing: [{1, 12, 13}] incoming: [{1, 12, 13}]
0 -> 2 = outgoing: [{0, 15}] incoming: [{0, 15}]
1 -> 0 = outgoing: [{8, 10, 3}] incoming: [{8, 10, 3}]
1 -> 2 = outgoing: [{0, 8, 15}] incoming: [{0, 8, 15}]
2 -> 0 = outgoing: [{10, 5}] incoming: [{10, 5}]
2 -> 1 = outgoing: [{15, 6, 7}] incoming: [{15, 6, 7}]
Notice that before on all nodes connections targeting the same node were created on the same shards, they would end up being processed by the same shards on the target node
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This would cause all the connections comming to node being handled on the same shard
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
makes sense.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
test approval
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
test reapproval
rpk/container: use 9093 for the external port for the kafka api
As pointed out in redpanda-data#8, I had no crc validation on decode. The most obvious place to add that validation (and length validation) is in kmsg, which also makes this validation more broadly applicable to those that do not want to use the kgo package. This has been tested with the chaining tests, and v0 / v1 message sets have been tested with kcl.
Regarding the test `serde_test` in `src/v/storage/tests/index_state_test.cc` $> ninja storage_log_index_rpunit $> bin/storage_log_index_rpunit Running 7 test cases... AddressSanitizer:DEADLYSIGNAL ================================================================= ==934285==ERROR: AddressSanitizer: stack-overflow on address 0x7fff2a209f58 (pc 0x558e9c2eab0e bp 0x7fff2a20a7a0 sp 0x7fff2a209f60 T0) #0 0x558e9c2eab0e in __asan_memmove /home/nwatkins/src/redpanda/vbuild/llvm/src/compiler-rt/lib/asan/asan_interceptors_memintrinsics.cpp:30:3 #1 0x558e9c448ec2 in std::__1::enable_if<(is_same<std::__1::remove_const<char const>::type, char>::value) && (is_trivially_copy_assignable<char>::value), char*>::type std::__1::__copy<char const, char>(char const*, char const*, char*) /home/nwatkins/src/redpanda/vbuild/llvm/in stall/bin/../include/c++/v1/algorithm:1735:9 #2 0x558e9c448e1e in char* std::__1::copy<char const*, char*>(char const*, char const*, char*) /home/nwatkins/src/redpanda/vbuild/llvm/install/bin/../include/c++/v1/algorithm:1748:16 #3 0x558e9c448c9a in std::__1::enable_if<__is_cpp17_random_access_iterator<char const*>::value, char*>::type std::__1::copy_n<char const*, unsigned long, char*>(char const*, unsigned long, char*) /home/nwatkins/src/redpanda/vbuild/llvm/install/bin/../include/c++/v1/algorithm:1 867:12 #4 0x558e9c674104 in iobuf_copy(details::io_iterator_consumer&, unsigned long)::$_4::operator()(char const*, unsigned long) const /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/bytes/iobuf.cc:145:13 #5 0x558e9c65aa69 in unsigned long details::io_iterator_consumer::consume<iobuf_copy(details::io_iterator_consumer&, unsigned long)::$_4>(unsigned long, iobuf_copy(details::io_iterator_consumer&, unsigned long)::$_4&&) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../sr c/v/bytes/details/io_iterator_consumer.h:141:45 #6 0x558e9c659af1 in iobuf_copy(details::io_iterator_consumer&, unsigned long) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/bytes/iobuf.cc:143:12 #7 0x558e9c57eeb3 in iobuf_parser_base::peek(unsigned long) const /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/bytes/iobuf_parser.h:100:16 #8 0x558e9c5751b6 in serde::peek_version(iobuf_parser&) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/serde/serde.h:547:43 #9 0x558e9c55ae54 in storage::read_nested(iobuf_parser&, storage::index_state&, unsigned long) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/storage/index_state.cc:257:9 #10 0x558e9c475297 in std::__1::decay<storage::index_state>::type serde::read_nested<storage::index_state>(iobuf_parser&, unsigned long) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/serde/serde.h:475:5 #11 0x558e9c55aedb in storage::read_nested(iobuf_parser&, storage::index_state&, unsigned long) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/storage/index_state.cc:258:14 #12 0x558e9c475297 in std::__1::decay<storage::index_state>::type serde::read_nested<storage::index_state>(iobuf_parser&, unsigned long) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/serde/serde.h:475:5 #13 0x558e9c55aedb in storage::read_nested(iobuf_parser&, storage::index_state&, unsigned long) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/storage/index_state.cc:258:14 #14 0x558e9c475297 in std::__1::decay<storage::index_state>::type serde::read_nested<storage::index_state>(iobuf_parser&, unsigned long) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/serde/serde.h:475:5 #15 0x558e9c55aedb in storage::read_nested(iobuf_parser&, storage::index_state&, unsigned long) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/storage/index_state.cc:258:14 ... === This makes the recursion go away, but the deserialized output is all zeros. So that probably isn't the solution. diff --git a/src/v/storage/index_state.cc b/src/v/storage/index_state.cc index 57808eae2..4aceb904f 100644 --- a/src/v/storage/index_state.cc +++ b/src/v/storage/index_state.cc @@ -255,7 +255,7 @@ iobuf index_state::checksum_and_serialize() { void read_nested(iobuf_parser& in, index_state& st, size_t bytes_left_limit) { if (serde::peek_version(in) > index_state::ondisk_version) { - st = serde::read_nested<index_state>(in, bytes_left_limit); + serde::read_nested<index_state>(in, st, bytes_left_limit); return; } Signed-off-by: Noah Watkins <[email protected]>
Regarding the test `serde_test` in `src/v/storage/tests/index_state_test.cc` $> ninja storage_log_index_rpunit $> bin/storage_log_index_rpunit Running 7 test cases... AddressSanitizer:DEADLYSIGNAL ================================================================= ==934285==ERROR: AddressSanitizer: stack-overflow on address 0x7fff2a209f58 (pc 0x558e9c2eab0e bp 0x7fff2a20a7a0 sp 0x7fff2a209f60 T0) #0 0x558e9c2eab0e in __asan_memmove /home/nwatkins/src/redpanda/vbuild/llvm/src/compiler-rt/lib/asan/asan_interceptors_memintrinsics.cpp:30:3 #1 0x558e9c448ec2 in std::__1::enable_if<(is_same<std::__1::remove_const<char const>::type, char>::value) && (is_trivially_copy_assignable<char>::value), char*>::type std::__1::__copy<char const, char>(char const*, char const*, char*) /home/nwatkins/src/redpanda/vbuild/llvm/in stall/bin/../include/c++/v1/algorithm:1735:9 #2 0x558e9c448e1e in char* std::__1::copy<char const*, char*>(char const*, char const*, char*) /home/nwatkins/src/redpanda/vbuild/llvm/install/bin/../include/c++/v1/algorithm:1748:16 #3 0x558e9c448c9a in std::__1::enable_if<__is_cpp17_random_access_iterator<char const*>::value, char*>::type std::__1::copy_n<char const*, unsigned long, char*>(char const*, unsigned long, char*) /home/nwatkins/src/redpanda/vbuild/llvm/install/bin/../include/c++/v1/algorithm:1 867:12 #4 0x558e9c674104 in iobuf_copy(details::io_iterator_consumer&, unsigned long)::$_4::operator()(char const*, unsigned long) const /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/bytes/iobuf.cc:145:13 #5 0x558e9c65aa69 in unsigned long details::io_iterator_consumer::consume<iobuf_copy(details::io_iterator_consumer&, unsigned long)::$_4>(unsigned long, iobuf_copy(details::io_iterator_consumer&, unsigned long)::$_4&&) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../sr c/v/bytes/details/io_iterator_consumer.h:141:45 #6 0x558e9c659af1 in iobuf_copy(details::io_iterator_consumer&, unsigned long) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/bytes/iobuf.cc:143:12 #7 0x558e9c57eeb3 in iobuf_parser_base::peek(unsigned long) const /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/bytes/iobuf_parser.h:100:16 #8 0x558e9c5751b6 in serde::peek_version(iobuf_parser&) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/serde/serde.h:547:43 #9 0x558e9c55ae54 in storage::read_nested(iobuf_parser&, storage::index_state&, unsigned long) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/storage/index_state.cc:257:9 #10 0x558e9c475297 in std::__1::decay<storage::index_state>::type serde::read_nested<storage::index_state>(iobuf_parser&, unsigned long) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/serde/serde.h:475:5 #11 0x558e9c55aedb in storage::read_nested(iobuf_parser&, storage::index_state&, unsigned long) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/storage/index_state.cc:258:14 #12 0x558e9c475297 in std::__1::decay<storage::index_state>::type serde::read_nested<storage::index_state>(iobuf_parser&, unsigned long) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/serde/serde.h:475:5 #13 0x558e9c55aedb in storage::read_nested(iobuf_parser&, storage::index_state&, unsigned long) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/storage/index_state.cc:258:14 #14 0x558e9c475297 in std::__1::decay<storage::index_state>::type serde::read_nested<storage::index_state>(iobuf_parser&, unsigned long) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/serde/serde.h:475:5 #15 0x558e9c55aedb in storage::read_nested(iobuf_parser&, storage::index_state&, unsigned long) /home/nwatkins/src/redpanda/vbuild/debug/clang/../../../src/v/storage/index_state.cc:258:14 ... === This makes the recursion go away, but the deserialized output is all zeros. So that probably isn't the solution. diff --git a/src/v/storage/index_state.cc b/src/v/storage/index_state.cc index 57808eae2..4aceb904f 100644 --- a/src/v/storage/index_state.cc +++ b/src/v/storage/index_state.cc @@ -255,7 +255,7 @@ iobuf index_state::checksum_and_serialize() { void read_nested(iobuf_parser& in, index_state& st, size_t bytes_left_limit) { if (serde::peek_version(in) > index_state::ondisk_version) { - st = serde::read_nested<index_state>(in, bytes_left_limit); + serde::read_nested<index_state>(in, st, bytes_left_limit); return; } Signed-off-by: Noah Watkins <[email protected]>
Without this change there is a data race that can result in a heap-use-after-free. ``` ==3190349==ERROR: AddressSanitizer: heap-use-after-free on address 0x60b000257420 at pc 0x14995faf26e4 bp 0x7ffd3e04b710 sp 0x7ffd3e04b708 READ of size 8 at 0x60b000257420 thread T0 #0 0x14995faf26e3 in profile_flush_file_data /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/util/profile/prof_file.c:468:24 #1 0x14995faf0cef in profile_close_file /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/util/profile/prof_file.c:559:14 #2 0x14995fb0c086 in profile_release /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/util/profile/prof_init.c:514:13 #3 0x14995fa924c6 in k5_os_free_context /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/lib/krb5/os/init_os_ctx.c:506:9 #4 0x14995f938b87 in krb5_free_context /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/lib/krb5/krb/init_ctx.c:294:5 #5 0x14995ffc41c3 in krb5_gss_delete_sec_context /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/lib/gssapi/krb5/delete_sec_context.c:87:9 #6 0x14995ff399a6 in gssint_delete_internal_sec_context /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/lib/gssapi/mechglue/g_glue.c:606:15 #7 0x14995ff25450 in gss_delete_sec_context /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/lib/gssapi/mechglue/g_delete_sec_context.c:91:11 #8 0x55e258b9c641 in security::gss::ctx_id::reset() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/security/gssapi.h:170:13 #9 0x55e258b9c408 in security::gss::ctx_id::~ctx_id() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/security/gssapi.h:165:17 #10 0x55e258b9c37c in security::gssapi_authenticator::impl::~impl() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/security/gssapi_authenticator.cc:116:29 #11 0x55e258b9c23c in std::__1::default_delete<security::gssapi_authenticator::impl>::operator()(security::gssapi_authenticator::impl*) const /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/llvm/install/bin/../include/c++/v1/__memory/unique_ptr.h:54:5 #12 0x55e258b65324 in std::__1::unique_ptr<security::gssapi_authenticator::impl, std::__1::default_delete<security::gssapi_authenticator::impl> >::reset(security::gssapi_authenticator::impl*) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/llvm/install/bin/../include/c++/v1/__memory/unique_ptr.h:315:7 #13 0x55e258b5f302 in security::gssapi_authenticator::authenticate(seastar::basic_sstring<unsigned char, unsigned int, 31u, false>) (.resume) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/security/gssapi_authenticator.cc:202:15 #14 0x55e258b11830 in std::__1::coroutine_handle<seastar::internal::coroutine_traits_base<boost::outcome_v2::basic_result<seastar::basic_sstring<unsigned char, unsigned int, 31u, false>, std::__1::error_code, boost::outcome_v2::policy::error_code_throw_as_system_error<seastar::basic_sstring<unsigned char, unsigned int, 31u, false>, std::__1::error_code, void> > >::promise_type>::resume() const /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/llvm/install/bin/../include/c++/v1/__coroutine/coroutine_handle.h:168:9 #15 0x55e258b11315 in seastar::internal::coroutine_traits_base<boost::outcome_v2::basic_result<seastar::basic_sstring<unsigned char, unsigned int, 31u, false>, std::__1::error_code, boost::outcome_v2::policy::error_code_throw_as_system_error<seastar::basic_sstring<unsigned char, unsigned int, 31u, false>, std::__1::error_code, void> > >::promise_type::run_and_dispose() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/rp_deps_install/include/seastar/core/coroutine.hh:78:20 #16 0x55e2594a1ead in seastar::reactor::run_tasks(seastar::reactor::task_queue&) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/seastar-prefix/src/seastar/src/core/reactor.cc:2330:14 #17 0x55e2594a7d0e in seastar::reactor::run_some_tasks() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/seastar-prefix/src/seastar/src/core/reactor.cc:2737:9 #18 0x55e2594ac86f in seastar::reactor::do_run() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/seastar-prefix/src/seastar/src/core/reactor.cc:2906:9 #19 0x55e2594aa3f8 in seastar::reactor::run() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/seastar-prefix/src/seastar/src/core/reactor.cc:2789:16 #20 0x55e2591c0a7c in seastar::app_template::run_deprecated(int, char**, std::__1::function<void ()>&&) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/seastar-prefix/src/seastar/src/core/app-template.cc:265:31 #21 0x55e2591be135 in seastar::app_template::run(int, char**, std::__1::function<seastar::future<int> ()>&&) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/seastar-prefix/src/seastar/src/core/app-template.cc:156:12 #22 0x55e2470c46e4 in application::run(int, char**) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/redpanda/application.cc:323:16 #23 0x55e247081d16 in main /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/redpanda/main.cc:22:16 #24 0x14995cc29d8f in __libc_start_call_main csu/../sysdeps/nptl/libc_start_call_main.h:58:16 #25 0x14995cc29e3f in __libc_start_main csu/../csu/libc-start.c:392:3 #26 0x55e246fc1844 in _start (/home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/bin/redpanda+0x3a678844) (BuildId: 0a81be2927716d279ef7fc96d20cc5d5dfbd1cb2) 0x60b000257420 is located 0 bytes inside of 104-byte region [0x60b000257420,0x60b000257488) freed by thread T0 here: #0 0x55e2470443e2 in free /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/llvm/src/compiler-rt/lib/asan/asan_malloc_linux.cpp:52:3 #1 0x14995faf3ed4 in profile_free_file_data /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/util/profile/prof_file.c:552:5 #2 0x14995faf3860 in profile_dereference_data_locked /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/util/profile/prof_file.c:509:9 #3 0x14995faf0b74 in profile_dereference_data /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/util/profile/prof_file.c:502:5 #4 0x14995faf3fc1 in profile_free_file /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/util/profile/prof_file.c:523:5 #5 0x14995faf0d38 in profile_close_file /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/util/profile/prof_file.c:562:5 #6 0x14995fb0c086 in profile_release /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/util/profile/prof_init.c:514:13 #7 0x14995fa924c6 in k5_os_free_context /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/lib/krb5/os/init_os_ctx.c:506:9 #8 0x14995f938b87 in krb5_free_context /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/lib/krb5/krb/init_ctx.c:294:5 #9 0x14995ffc41c3 in krb5_gss_delete_sec_context /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/lib/gssapi/krb5/delete_sec_context.c:87:9 #10 0x14995ff399a6 in gssint_delete_internal_sec_context /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/lib/gssapi/mechglue/g_glue.c:606:15 #11 0x14995ff25450 in gss_delete_sec_context /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/lib/gssapi/mechglue/g_delete_sec_context.c:91:11 #12 0x55e258b9c641 in security::gss::ctx_id::reset() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/security/gssapi.h:170:13 #13 0x55e258b9c408 in security::gss::ctx_id::~ctx_id() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/security/gssapi.h:165:17 #14 0x55e258b9c37c in security::gssapi_authenticator::impl::~impl() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/security/gssapi_authenticator.cc:116:29 #15 0x55e258b9c23c in std::__1::default_delete<security::gssapi_authenticator::impl>::operator()(security::gssapi_authenticator::impl*) const /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/llvm/install/bin/../include/c++/v1/__memory/unique_ptr.h:54:5 #16 0x55e258b65324 in std::__1::unique_ptr<security::gssapi_authenticator::impl, std::__1::default_delete<security::gssapi_authenticator::impl> >::reset(security::gssapi_authenticator::impl*) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/llvm/install/bin/../include/c++/v1/__memory/unique_ptr.h:315:7 #17 0x55e258b5f302 in security::gssapi_authenticator::authenticate(seastar::basic_sstring<unsigned char, unsigned int, 31u, false>) (.resume) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/security/gssapi_authenticator.cc:202:15 #18 0x55e258b11830 in std::__1::coroutine_handle<seastar::internal::coroutine_traits_base<boost::outcome_v2::basic_result<seastar::basic_sstring<unsigned char, unsigned int, 31u, false>, std::__1::error_code, boost::outcome_v2::policy::error_code_throw_as_system_error<seastar::basic_sstring<unsigned char, unsigned int, 31u, false>, std::__1::error_code, void> > >::promise_type>::resume() const /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/llvm/install/bin/../include/c++/v1/__coroutine/coroutine_handle.h:168:9 #19 0x55e258b11315 in seastar::internal::coroutine_traits_base<boost::outcome_v2::basic_result<seastar::basic_sstring<unsigned char, unsigned int, 31u, false>, std::__1::error_code, boost::outcome_v2::policy::error_code_throw_as_system_error<seastar::basic_sstring<unsigned char, unsigned int, 31u, false>, std::__1::error_code, void> > >::promise_type::run_and_dispose() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/rp_deps_install/include/seastar/core/coroutine.hh:78:20 #20 0x55e2594a1ead in seastar::reactor::run_tasks(seastar::reactor::task_queue&) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/seastar-prefix/src/seastar/src/core/reactor.cc:2330:14 #21 0x55e2594a7d0e in seastar::reactor::run_some_tasks() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/seastar-prefix/src/seastar/src/core/reactor.cc:2737:9 #22 0x55e2594ac86f in seastar::reactor::do_run() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/seastar-prefix/src/seastar/src/core/reactor.cc:2906:9 #23 0x55e2594aa3f8 in seastar::reactor::run() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/seastar-prefix/src/seastar/src/core/reactor.cc:2789:16 #24 0x55e2591c0a7c in seastar::app_template::run_deprecated(int, char**, std::__1::function<void ()>&&) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/seastar-prefix/src/seastar/src/core/app-template.cc:265:31 #25 0x55e2591be135 in seastar::app_template::run(int, char**, std::__1::function<seastar::future<int> ()>&&) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/seastar-prefix/src/seastar/src/core/app-template.cc:156:12 #26 0x55e2470c46e4 in application::run(int, char**) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/redpanda/application.cc:323:16 #27 0x55e247081d16 in main /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/redpanda/main.cc:22:16 #28 0x14995cc29d8f in __libc_start_call_main csu/../sysdeps/nptl/libc_start_call_main.h:58:16 previously allocated by thread T4 here: #0 0x55e24704468e in malloc /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/llvm/src/compiler-rt/lib/asan/asan_malloc_linux.cpp:69:3 #1 0x14995faee48d in profile_make_prf_data /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/util/profile/prof_file.c:150:9 #2 0x14995faf0360 in profile_open_file /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/util/profile/prof_file.c:233:12 #3 0x14995fb0ab1e in profile_init_flags /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/util/profile/prof_init.c:190:22 #4 0x14995fa90bc0 in os_init_paths /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/lib/krb5/os/init_os_ctx.c:387:18 #5 0x14995fa9090e in k5_os_init_context /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/lib/krb5/os/init_os_ctx.c:438:18 #6 0x14995f936ece in krb5_init_context_profile /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/lib/krb5/krb/init_ctx.c:209:14 #7 0x14995f936871 in krb5_init_context /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/lib/krb5/krb/init_ctx.c:139:12 #8 0x14995fffc70b in krb5_gss_init_context /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/lib/gssapi/krb5/init_sec_context.c:1047:12 #9 0x14995ffa0719 in kg_accept_krb5 /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/lib/gssapi/krb5/accept_sec_context.c:694:12 #10 0x14995ff9e686 in krb5_gss_accept_sec_context_ext /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/lib/gssapi/krb5/accept_sec_context.c:1311:12 #11 0x14995ffaaf9c in krb5_gss_accept_sec_context /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/lib/gssapi/krb5/accept_sec_context.c:1340:12 #12 0x14995ff05dbd in gss_accept_sec_context /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/lib/gssapi/mechglue/g_accept_sec_context.c:266:15 #13 0x55e258b2b157 in security::gssapi_authenticator::impl::more(std::__1::basic_string_view<unsigned char, std::__1::char_traits<unsigned char> >) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/security/gssapi_authenticator.cc:292:25 #14 0x55e258b2946c in security::gssapi_authenticator::impl::authenticate(seastar::basic_sstring<unsigned char, unsigned int, 31u, false>) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/security/gssapi_authenticator.cc:218:16 #15 0x55e258b3b562 in security::gssapi_authenticator::authenticate(seastar::basic_sstring<unsigned char, unsigned int, 31u, false>)::$_0::operator()() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/security/gssapi_authenticator.cc:195:25 #16 0x55e258b3a96a in ssx::impl::worker_task<security::gssapi_authenticator::authenticate(seastar::basic_sstring<unsigned char, unsigned int, 31u, false>)::$_0>::process(seastar::alien::instance&, unsigned int) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/ssx/thread_worker.h:67:37 #17 0x55e24764d1bc in ssx::impl::thread_worker::run() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/ssx/thread_worker.h:159:20 #18 0x55e24764ca9f in ssx::impl::thread_worker::start()::'lambda'()::operator()() const /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/ssx/thread_worker.h:104:49 #19 0x55e24764c910 in decltype(static_cast<ssx::impl::thread_worker::start()::'lambda'()>(fp)()) std::__1::__invoke<ssx::impl::thread_worker::start()::'lambda'()>(ssx::impl::thread_worker::start()::'lambda'()&&) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/llvm/install/bin/../include/c++/v1/type_traits:3640:23 #20 0x55e24764c808 in void std::__1::__thread_execute<std::__1::unique_ptr<std::__1::__thread_struct, std::__1::default_delete<std::__1::__thread_struct> >, ssx::impl::thread_worker::start()::'lambda'()>(std::__1::tuple<std::__1::unique_ptr<std::__1::__thread_struct, std::__1::default_delete<std::__1::__thread_struct> >, ssx::impl::thread_worker::start()::'lambda'()>&, std::__1::__tuple_indices<>) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/llvm/install/bin/../include/c++/v1/thread:282:5 #21 0x55e24764ad39 in void* std::__1::__thread_proxy<std::__1::tuple<std::__1::unique_ptr<std::__1::__thread_struct, std::__1::default_delete<std::__1::__thread_struct> >, ssx::impl::thread_worker::start()::'lambda'()> >(void*) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/llvm/install/bin/../include/c++/v1/thread:293:5 #22 0x14995cc94b42 in start_thread nptl/./nptl/pthread_create.c:442:8 Thread T4 created by T0 here: #0 0x55e24702db0c in pthread_create /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/llvm/src/compiler-rt/lib/asan/asan_interceptors.cpp:208:3 #1 0x55e24764ab0c in std::__1::__libcpp_thread_create(unsigned long*, void* (*)(void*), void*) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/llvm/install/bin/../include/c++/v1/__threading_support:375:10 #2 0x55e24764a106 in std::__1::thread::thread<ssx::impl::thread_worker::start()::'lambda'(), void>(ssx::impl::thread_worker::start()::'lambda'()&&) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/llvm/install/bin/../include/c++/v1/thread:309:16 #3 0x55e247606d40 in ssx::impl::thread_worker::start() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/ssx/thread_worker.h:104:19 #4 0x55e247305329 in ssx::thread_worker::start() (.resume) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/ssx/thread_worker.h:199:15 #5 0x55e24756b680 in std::__1::coroutine_handle<seastar::internal::coroutine_traits_base<void>::promise_type>::resume() const /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/llvm/install/bin/../include/c++/v1/__coroutine/coroutine_handle.h:168:9 #6 0x55e24756b1a5 in seastar::internal::coroutine_traits_base<void>::promise_type::run_and_dispose() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/rp_deps_install/include/seastar/core/coroutine.hh:120:20 #7 0x55e2594a1ead in seastar::reactor::run_tasks(seastar::reactor::task_queue&) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/seastar-prefix/src/seastar/src/core/reactor.cc:2330:14 #8 0x55e2594a7d0e in seastar::reactor::run_some_tasks() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/seastar-prefix/src/seastar/src/core/reactor.cc:2737:9 #9 0x55e2594ac86f in seastar::reactor::do_run() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/seastar-prefix/src/seastar/src/core/reactor.cc:2906:9 #10 0x55e2594aa3f8 in seastar::reactor::run() /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/seastar-prefix/src/seastar/src/core/reactor.cc:2789:16 #11 0x55e2591c0a7c in seastar::app_template::run_deprecated(int, char**, std::__1::function<void ()>&&) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/seastar-prefix/src/seastar/src/core/app-template.cc:265:31 #12 0x55e2591be135 in seastar::app_template::run(int, char**, std::__1::function<seastar::future<int> ()>&&) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/seastar-prefix/src/seastar/src/core/app-template.cc:156:12 #13 0x55e2470c46e4 in application::run(int, char**) /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/redpanda/application.cc:323:16 #14 0x55e247081d16 in main /home/ben/development/src/github.com/BenPope/redpanda-clang-13/src/v/redpanda/main.cc:22:16 #15 0x14995cc29d8f in __libc_start_call_main csu/../sysdeps/nptl/libc_start_call_main.h:58:16 SUMMARY: AddressSanitizer: heap-use-after-free /home/ben/development/src/github.com/BenPope/redpanda-clang-13/vbuild/debug/clang/v_deps_build/krb5-prefix/src/krb5/src/util/profile/prof_file.c:468:24 in profile_flush_file_data Shadow bytes around the buggy address: 0x0c1680042e30: fd fd fd fd fd fd fd fd fd fd fa fa fa fa fa fa 0x0c1680042e40: fa fa fd fd fd fd fd fd fd fd fd fd fd fd fd fa 0x0c1680042e50: fa fa fa fa fa fa fa fa fd fd fd fd fd fd fd fd 0x0c1680042e60: fd fd fd fd fd fd fa fa fa fa fa fa fa fa fd fd 0x0c1680042e70: fd fd fd fd fd fd fd fd fd fd fd fd fa fa fa fa =>0x0c1680042e80: fa fa fa fa[fd]fd fd fd fd fd fd fd fd fd fd fd 0x0c1680042e90: fd fa fa fa fa fa fa fa fa fa fd fd fd fd fd fd 0x0c1680042ea0: fd fd fd fd fd fd fd fd fa fa fa fa fa fa fa fa 0x0c1680042eb0: fd fd fd fd fd fd fd fd fd fd fd fd fd fa fa fa 0x0c1680042ec0: fa fa fa fa fa fa fd fd fd fd fd fd fd fd fd fd 0x0c1680042ed0: fd fd fd fd fa fa fa fa fa fa fa fa fd fd fd fd Shadow byte legend (one shadow byte represents 8 application bytes): Addressable: 00 Partially addressable: 01 02 03 04 05 06 07 Heap left redzone: fa Freed heap region: fd Stack left redzone: f1 Stack mid redzone: f2 Stack right redzone: f3 Stack after return: f5 Stack use after scope: f8 Global redzone: f9 Global init order: f6 Poisoned by user: f7 Container overflow: fc Array cookie: ac Intra object redzone: bb ASan internal: fe Left alloca redzone: ca Right alloca redzone: cb ==3190349==ABORTING ``` Signed-off-by: Ben Pope <[email protected]>
This segfault is triggering as the test does something innocuous like logging a few string: ``` #0 0x7f875b0bbc1f in std::__1::basic_ostream<char, std::__1::char_traits<char> >::sentry::sentry(std::__1::basic_ostream<char, std::__1::char_traits<char> >&) (/vectorized/lib/libc++.so.1+0x65c1f) redpanda-data#1 0x5597c3979afc in std::__1::basic_ostream<char, std::__1::char_traits<char> >& std::__1::__put_character_sequence<char, std::__1::char_traits<char> >(std::__1::basic_ostream<char, std::__1::char_traits<char> >&, char const*, unsigned long) /vectorized/llvm/bin/../include/c++/v1/ostream:722:57 redpanda-data#2 0x5597c4079e44 in std::__1::basic_ostream<char, std::__1::char_traits<char> >& std::__1::operator<<<char, std::__1::char_traits<char> >(std::__1::basic_ostream<char, std::__1::char_traits<char> >&, std::__1::basic_string_view<char, std::__1::char_traits<char> >) /vectorized/llvm/bin/../include/c++/v1/ostream:1057:12 redpanda-data#3 0x5597c602c9f3 in seastar::logger::do_log(seastar::log_level, seastar::logger::log_writer&) /v/build/v_deps_build/seastar-prefix/src/seastar/src/util/log.cc:342:15 redpanda-data#4 0x5597c3c4ad20 in void seastar::logger::log<char const*, int, seastar::basic_sstring<char, unsigned int, 15u, true>&>(seastar::log_level, seastar::logger::format_info, char const*&&, int&&, seastar::basic_sstring<char, unsigned int, 15u, true>&) /vectorized/include/seastar/util/log.hh:231:17 redpanda-data#5 0x5597c3f23c51 in void seastar::logger::trace<char const*, int, seastar::basic_sstring<char, unsigned int, 15u, true>&>(seastar::logger::format_info, char const*&&, int&&, seastar::basic_sstring<char, unsigned int, 15u, true>&) /vectorized/include/seastar/util/log.hh:385:9 redpanda-data#6 0x5597c3f14b95 in cloud_roles::signature_v4::sign_header(boost::beast::http::header<true, boost::beast::http::basic_fields<std::__1::allocator<char> > >&, std::__1::basic_string_view<char, std::__1::char_traits<char> >) const /var/lib/buildkite-agent/builds/buildkite-amd64-builders-i-00e717c5e253e2891-1/redpanda/vtools/src/v/cloud_roles/signature.cc:382:5 redpanda-data#7 0x5597c3bf1dfb in cloud_roles::apply_aws_credentials::add_auth(boost::beast::http::header<true, boost::beast::http::basic_fields<std::__1::allocator<char> > >&) const /var/lib/buildkite-agent/builds/buildkite-amd64-builders-i-00e717c5e253e2891-1/redpanda/vtools/src/v/cloud_roles/apply_aws_credentials.cc:83:23 redpanda-data#8 0x5597c3b8f88d in cloud_roles::apply_credentials::add_auth(boost::beast::http::header<true, boost::beast::http::basic_fields<std::__1::allocator<char> > >&) const /var/lib/buildkite-agent/builds/buildkite-amd64-builders-i-00e717c5e253e2891-1/redpanda/vtools/src/v/cloud_roles/apply_credentials.h:47:23 redpanda-data#9 0x5597c3b8c578 in test_aws_headers::test_method() /var/lib/buildkite-agent/builds/buildkite-amd64-builders-i-00e717c5e253e2891-1/redpanda/vtools/src/v/cloud_roles/tests/credential_tests.cc:50:17 ``` What is suspicious is that this test is using seastar's logger from a non-seastar reactor thread. Taking a look we do see some things like this_shard_id() being called, but other than that it doesn't look like it should be problematic. But, it's unclear if there is more lurking there. Changed the test to run things that log using seastar logger on a seastar thread. Signed-off-by: Noah Watkins <[email protected]>
The tee_wrapper helper was being used to set the ostream pointer for global logger to be an ostream instance that lived only on the stack, and not resetting it to something global like std::cerr after the test finished. This resulted in the following segfault in other tests that ran after the change but didn't make their own change to ostream pointer: ``` #0 0x7f875b0bbc1f in std::__1::basic_ostream<char, std::__1::char_traits<char> >::sentry::sentry(std::__1::basic_ostream<char, std::__1::char_traits<char> >&) (/vectorized/lib/libc++.so.1+0x65c1f) #1 0x5597c3979afc in std::__1::basic_ostream<char, std::__1::char_traits<char> >& std::__1::__put_character_sequence<char, std::__1::char_traits<char> >(std::__1::basic_ostream<char, std::__1::char_traits<char> >&, char const*, unsigned long) /vectorized/llvm/bin/../include/c++/v1/ostream:722:57 #2 0x5597c4079e44 in std::__1::basic_ostream<char, std::__1::char_traits<char> >& std::__1::operator<<<char, std::__1::char_traits<char> >(std::__1::basic_ostream<char, std::__1::char_traits<char> >&, std::__1::basic_string_view<char, std::__1::char_traits<char> >) /vectorized/llvm/bin/../include/c++/v1/ostream:1057:12 #3 0x5597c602c9f3 in seastar::logger::do_log(seastar::log_level, seastar::logger::log_writer&) /v/build/v_deps_build/seastar-prefix/src/seastar/src/util/log.cc:342:15 #4 0x5597c3c4ad20 in void seastar::logger::log<char const*, int, seastar::basic_sstring<char, unsigned int, 15u, true>&>(seastar::log_level, seastar::logger::format_info, char const*&&, int&&, seastar::basic_sstring<char, unsigned int, 15u, true>&) /vectorized/include/seastar/util/log.hh:231:17 #5 0x5597c3f23c51 in void seastar::logger::trace<char const*, int, seastar::basic_sstring<char, unsigned int, 15u, true>&>(seastar::logger::format_info, char const*&&, int&&, seastar::basic_sstring<char, unsigned int, 15u, true>&) /vectorized/include/seastar/util/log.hh:385:9 #6 0x5597c3f14b95 in cloud_roles::signature_v4::sign_header(boost::beast::http::header<true, boost::beast::http::basic_fields<std::__1::allocator<char> > >&, std::__1::basic_string_view<char, std::__1::char_traits<char> >) const /var/lib/buildkite-agent/builds/buildkite-amd64-builders-i-00e717c5e253e2891-1/redpanda/vtools/src/v/cloud_roles/signature.cc:382:5 #7 0x5597c3bf1dfb in cloud_roles::apply_aws_credentials::add_auth(boost::beast::http::header<true, boost::beast::http::basic_fields<std::__1::allocator<char> > >&) const /var/lib/buildkite-agent/builds/buildkite-amd64-builders-i-00e717c5e253e2891-1/redpanda/vtools/src/v/cloud_roles/apply_aws_credentials.cc:83:23 #8 0x5597c3b8f88d in cloud_roles::apply_credentials::add_auth(boost::beast::http::header<true, boost::beast::http::basic_fields<std::__1::allocator<char> > >&) const /var/lib/buildkite-agent/builds/buildkite-amd64-builders-i-00e717c5e253e2891-1/redpanda/vtools/src/v/cloud_roles/apply_credentials.h:47:23 #9 0x5597c3b8c578 in test_aws_headers::test_method() /var/lib/buildkite-agent/builds/buildkite-amd64-builders-i-00e717c5e253e2891-1/redpanda/vtools/src/v/cloud_roles/tests/credential_tests.cc:50:17 ``` This change resets the ostream pointer in the wrapper destructor. Signed-off-by: Noah Watkins <[email protected]>
The tee_wrapper helper was being used to set the ostream pointer for global logger to be an ostream instance that lived only on the stack, and not resetting it to something global like std::cerr after the test finished. This resulted in the following segfault in other tests that ran after the change but didn't make their own change to ostream pointer: ``` #0 0x7f875b0bbc1f in std::__1::basic_ostream<char, std::__1::char_traits<char> >::sentry::sentry(std::__1::basic_ostream<char, std::__1::char_traits<char> >&) (/vectorized/lib/libc++.so.1+0x65c1f) #1 0x5597c3979afc in std::__1::basic_ostream<char, std::__1::char_traits<char> >& std::__1::__put_character_sequence<char, std::__1::char_traits<char> >(std::__1::basic_ostream<char, std::__1::char_traits<char> >&, char const*, unsigned long) /vectorized/llvm/bin/../include/c++/v1/ostream:722:57 redpanda-data#2 0x5597c4079e44 in std::__1::basic_ostream<char, std::__1::char_traits<char> >& std::__1::operator<<<char, std::__1::char_traits<char> >(std::__1::basic_ostream<char, std::__1::char_traits<char> >&, std::__1::basic_string_view<char, std::__1::char_traits<char> >) /vectorized/llvm/bin/../include/c++/v1/ostream:1057:12 redpanda-data#3 0x5597c602c9f3 in seastar::logger::do_log(seastar::log_level, seastar::logger::log_writer&) /v/build/v_deps_build/seastar-prefix/src/seastar/src/util/log.cc:342:15 redpanda-data#4 0x5597c3c4ad20 in void seastar::logger::log<char const*, int, seastar::basic_sstring<char, unsigned int, 15u, true>&>(seastar::log_level, seastar::logger::format_info, char const*&&, int&&, seastar::basic_sstring<char, unsigned int, 15u, true>&) /vectorized/include/seastar/util/log.hh:231:17 redpanda-data#5 0x5597c3f23c51 in void seastar::logger::trace<char const*, int, seastar::basic_sstring<char, unsigned int, 15u, true>&>(seastar::logger::format_info, char const*&&, int&&, seastar::basic_sstring<char, unsigned int, 15u, true>&) /vectorized/include/seastar/util/log.hh:385:9 redpanda-data#6 0x5597c3f14b95 in cloud_roles::signature_v4::sign_header(boost::beast::http::header<true, boost::beast::http::basic_fields<std::__1::allocator<char> > >&, std::__1::basic_string_view<char, std::__1::char_traits<char> >) const /var/lib/buildkite-agent/builds/buildkite-amd64-builders-i-00e717c5e253e2891-1/redpanda/vtools/src/v/cloud_roles/signature.cc:382:5 redpanda-data#7 0x5597c3bf1dfb in cloud_roles::apply_aws_credentials::add_auth(boost::beast::http::header<true, boost::beast::http::basic_fields<std::__1::allocator<char> > >&) const /var/lib/buildkite-agent/builds/buildkite-amd64-builders-i-00e717c5e253e2891-1/redpanda/vtools/src/v/cloud_roles/apply_aws_credentials.cc:83:23 redpanda-data#8 0x5597c3b8f88d in cloud_roles::apply_credentials::add_auth(boost::beast::http::header<true, boost::beast::http::basic_fields<std::__1::allocator<char> > >&) const /var/lib/buildkite-agent/builds/buildkite-amd64-builders-i-00e717c5e253e2891-1/redpanda/vtools/src/v/cloud_roles/apply_credentials.h:47:23 redpanda-data#9 0x5597c3b8c578 in test_aws_headers::test_method() /var/lib/buildkite-agent/builds/buildkite-amd64-builders-i-00e717c5e253e2891-1/redpanda/vtools/src/v/cloud_roles/tests/credential_tests.cc:50:17 ``` This change resets the ostream pointer in the wrapper destructor. Signed-off-by: Noah Watkins <[email protected]>
Made inter-node connections shard assignment deterministic.
Client side
Seastar in
posix-stack
for outgoing connections uses following port_assignment implementation:This algorithm makes outgoing port to be shard dependent.
Server side:
Incoming connections are assigned to shards using one of the following policies:
By default connections are distributed using simple round robin algorithm. In worst case scenario it can lead to the situation in which all active connections are handled on single cores. (Multiple re-connections are required).
We can leverage the fact that all the nodes in the cluster have unique ids and seastar outgoing port assignment being shard dependent to assign connection to shard deterministically on target node.
It is important to note that proposed solution is simple but it will provide perfect connection-shard assignment only on cluster with uniform number of shards per node.
Implemented solution
load_balancing_algorithm::port
on the server to assign connection to shard.Changes since v1:
boost::hash_combine
instead of simple modulo