thevar1able/ClickHouse
1
0
Fork 0
mirror of https://github.com/ClickHouse/ClickHouse.git synced 2024-12-03 13:02:00 +00:00
Code Issues Packages Projects Releases Wiki Activity
79,555 Commits 9,183 Branches 1,782 Tags 2.2 GiB
6aebc3e94c
Commit Graph

41 Commits

Author SHA1 Message Date
Azat Khuzhin
6aebc3e94c Do not loose ProfileEvents in case of thread destroyed before 
v2: drop std::move and add copy ctor for ProfileEvents::Counter::Snapshot
v2: remove std::move
 2021-12-19 23:43:05 +03:00
Azat Khuzhin
909ce68b5a Do not suppress exception in ThreadStatus::~ThreadStatus() 2021-12-19 23:43:05 +03:00
Alexey Milovidov
8b4a6a2416 Remove cruft 2021-10-28 02:10:39 +03:00
Nikita Mikhaylov
10487f75b2
Merge pull request #28364 from ClickHouse/profile-events-wip 
Forward profile events to client
 2021-10-12 13:16:01 +03:00
Dmitry Novik
9071a7151e Fix communication & race conditions 2021-10-11 17:39:23 +03:00
Dmitry Novik
356723427d WIP on ProfileEvents forwarding 2021-10-11 17:39:23 +03:00
Azat Khuzhin
9515430517 Fix compilation with glibc 2.34 (MINSIGSTKSZ defined as sysconf(_SC_SIGSTKSZ)) 
In glibc 2.34 MINSIGSTKSZ had been defined to sysconf(_SC_SIGSTKSZ) [1].

  [1]: https://sourceware.org/git/?p=glibc.git;a=commit;h=6c57d320484988e87e446e2e60ce42816bf51d53
 2021-10-07 00:18:56 +03:00
Alexey Milovidov
fe6b7c77c7 Rename "common" to "base" 2021-10-02 10:13:14 +03:00
Sergei Semin
f74350c148 fix special build 2021-09-12 02:49:31 +03:00
Raúl Marín
820a6e0987 Refactor MV stage metric setup and teardown 2021-08-10 10:49:12 +02:00
Raúl Marín
fc23e212c2 ThreadStatus: Don't change current thread if it's not this 2021-07-27 16:29:33 +02:00
Azat Khuzhin
3c76986311 Fix alternative stack under osx (MINSIGSTKSZ is 32K) 
Fixes: #25632
 2021-06-24 00:33:33 +03:00
alexey-milovidov
9a4125dd7c
Update ThreadStatus.cpp 2021-06-21 12:06:41 +03:00
Azat Khuzhin
12eadc2b5e Add a guard page for alternative signal handling stack 2021-06-20 20:08:50 +03:00
Azat Khuzhin
4dcbf3a5c5 Replace __pthread_get_minstack() with a const 2021-06-20 14:03:04 +03:00
Azat Khuzhin
9a3a1397e6 Fix alternative stack for SIGSEGV handling 
Previous stack size of 4096 wasn't enough, and this lead to that on
SIGSEGV signal handler may overwrite some TLS data (like in the
following example [1]) and in real binary "current_thread" was
overwritten, and this leads to lack of query-id on SIGSEGV, like here
[2].

  [1]: https://gist.github.com/azat/2ee360fd4f2828d363b0926431afacc6
  [2]: https://clickhouse-test-reports.s3.yandex.net/24411/79563953201ce6249a8fd49e22be3902ca4ee43a/fuzzer_ubsan/report.html#fail1

Fix this by using __pthread_get_minstack() that takes TLS block into
account. Yes it is private, but let's try (like rust tried).

This patch will also decrease TLS, since now the per-thread stack will
be allocated in the heap, not on the stack of each thread.

Refs: https://sourceware.org/legacy-ml/libc-alpha/2012-06/msg00335.html
Refs: https://maskray.me/blog/2021-02-14-all-about-thread-local-storage
Refs: https://sourceware.org/bugzilla/show_bug.cgi?id=11787
 2021-06-19 22:47:01 +03:00
Ivan
495c6e03aa
Replace all Context references with std::weak_ptr (#22297) 
* Replace all Context references with std::weak_ptr

* Fix shared context captured by value

* Fix build

* Fix Context with named sessions

* Fix copy context

* Fix gcc build

* Merge with master and fix build

* Fix gcc-9 build
 2021-04-11 02:33:54 +03:00
Alexander Tokmakov
004517009e fix 2021-01-29 18:13:27 +03:00
Alexander Tokmakov
ffaa8e34a6 minor code improvements around ThreadStatus 2021-01-28 16:57:36 +03:00
Azat Khuzhin
46c3266c6a Respect MINSIGSTKSZ for alternative stack to fix under aarch64 
And acording to sigaltstack(2) alignment is not required:

    When a signal handler is invoked on the alternate stack, the kernel
    automatically aligns the address given in ss.ss_sp to a suitable address
    boundary for the underlying hardware architecture.

Fixes: #18785
 2021-01-07 18:28:23 +03:00
Alexey Milovidov
be884a89f8 Minor fixes for min/sim hash 2020-12-29 13:16:43 +03:00
Alexey Milovidov
7f280bef37 Merge branch 'master' into sigaltstack 2020-12-17 05:09:10 +03:00
Alexander Kuzmenkov
b16c5a1748 Merge remote-tracking branch 'origin/master' into HEAD 2020-11-17 22:54:54 +03:00
Alexander Tokmakov
b94cc5c4e5 remove more stringstreams 2020-11-10 21:22:26 +03:00
Alexander Kuzmenkov
26229ed231 tmp spans for threads 
(doesn't compile because of json metadata changes)
 2020-11-09 18:07:38 +03:00
Alexey Milovidov
fd84d16387 Fix "server failed to start" error 2020-11-07 03:14:53 +03:00
alexey-milovidov
a591d1f32a
Update ThreadStatus.cpp 2020-10-26 22:46:25 +03:00
Alexey Milovidov
0ba2fa8105 Fix "Arcadia" 2020-10-26 11:56:30 +03:00
Alexey Milovidov
88e846b36d Fix build on Mac OS 2020-10-26 07:42:22 +03:00
Alexey Milovidov
068235c2d7 Disable alt stack for sanitizers 2020-10-26 05:18:36 +03:00
Alexey Milovidov
4e2c0ba226 Provide diagnostics on stack overflow 2020-10-25 03:10:05 +03:00
Azat Khuzhin
72d7b6117e Use total_memory_tracker when there is no other MemoryTracker object. 
This should significantly reduce the MemoryTracking drift, test shows
that there is 0 drift after query storm (100 queries, via http/tcp/tcp
in one session).

TL;DR;

To track memory, clickhouse creates memory tracker object for each
thread **explicitly**, but until it is not created the memory
allocations are not under account.
There should not be lot of allocations w/o memory tracker, since most of
the time it is created early enough, but even this maybe enough to
trigger some problems.

Plus sometimes it is not possible to create it, for example some 3d
party library does not allow to do this explicitly:
- for example before #15740 allocations from librdkafka threads,
- or even worse, poco threads, they don't have any routines to do this.
This won't be a problem for `MemoryTracking` metric if the deallocation
will be done from the same thread w/o memory tracker (or vise versa),
but this is not always true.

NOTE, that this will slow down per-thread allocations w/o memory
tracker, since before this patch there were no memory tracking for them
while now they will be accounted in total_memory_tracker, and for
total_memory_tracker max_untracked_memory is always reached.
But this should not be significant.
 2020-10-23 21:07:52 +03:00
Alexey Milovidov
31cbdd1a56 system.crash_log: development 2020-07-09 07:15:45 +03:00
Alexey Milovidov
7ba25bd799 Make it work 2020-06-20 14:17:15 +03:00
Alexander Kuzmenkov
6e3bbf83e3 thread-local perf events 2020-05-29 16:04:32 +03:00
Andrey Skobtsov
4816abed28 Settings for measuring only particular metrics via perf events 2020-05-20 21:25:49 +03:00
Andrey Skobtsov
cae41826a8 Merge branch 'master' into perf_events_metrics 
# Conflicts:
#	src/Common/ThreadProfileEvents.cpp
#	src/Common/ThreadProfileEvents.h
#	src/Common/ThreadStatus.cpp
#	src/Common/ThreadStatus.h
 2020-05-17 00:21:22 +03:00
Alexander Kazakov
e9baaa439b
Implementation of new system metrics provider (Procfs) (#10544) 
* New metrics provider (Procfs) + Refactored TasksStatsCounters

* Trivial statless test that ProcFS is provided

* Trivial perf test for ProcfsMetricsProvider

Co-authored-by: alexey-milovidov <milovidov@yandex-team.ru>
 2020-05-01 21:47:41 +03:00
Alexey Milovidov
be22a4b94e Checkpoint 2020-04-22 08:39:31 +03:00
Andrey Skobtsov
afaa7eca51 Merge branch 'master' into perf_events_metrics 
# Conflicts:
#	dbms/src/Common/ProfileEvents.cpp
 2020-04-13 23:59:48 +03:00
Ivan Lezhankin
06446b4f08 dbms/ → src/ 2020-04-03 18:14:31 +03:00