Consider nested cgroup memory limits for the total memory tracker

[rschu1ze]

Review note: Contains commits

• 92a00d3 - Cosmetics.
• ad0c0d8 - Cosmetics but also fixes the edge case that cgroup memory limit "0" was incorrectly interpreted as "no limit" but it really means "zero" memory.
• 3243ea0 Kills cgroups v1 supprt for simplicity. V1 has a different hierarchy model which would have made the next commit more complex. Also, v1 is dead since over half a decade.
• e997168 See the changelog message
• EDIT: 80a0e05 Resurrects cgroups v1. They are still enabled in many Linux distributions in GCP, AWS and Azure :-(
• ad13a57 Improves the detection of v2 features a bit

Changelog category (leave one):

• Improvement

Changelog entry (a user-readable short description of the changes that goes to CHANGELOG.md):

At startup and whenever config files are changed, ClickHouse updates the hard memory limits of its total memory tracker. These limits are computed based on various server settings and cgroups limits (on Linux). Previously, setting /sys/fs/cgroup/memory.max (for cgroups v2) was hard-coded. As a result, cgroup v2 memory limits configured for nested groups (hierarchies), e.g. /sys/fs/cgroup/my/nested/group/memory.max were ignored. This is now fixed. The behavior of v1 memory limits remains unchanged.

[robot-clickhouse-ci-1]

This is an automated comment for commit ad13a57 with description of existing statuses. It's updated for the latest CI running

✅ Click here to open a full report in a separate page

Successful checks

Check name	Description	Status
AST fuzzer	Runs randomly generated queries to catch program errors. The build type is optionally given in parenthesis. If it fails, ask a maintainer for help	✅ success
CI running	A meta-check that indicates the running CI. Normally, it's in success or pending state. The failed status indicates some problems with the PR	✅ success
ClickBench	Runs [ClickBench](https://github.com/ClickHouse/ClickBench/) with instant-attach table	✅ success
ClickHouse build check	Builds ClickHouse in various configurations for use in further steps. You have to fix the builds that fail. Build logs often has enough information to fix the error, but you might have to reproduce the failure locally. The cmake options can be found in the build log, grepping for cmake. Use these options and follow the general build process	✅ success
Compatibility check	Checks that clickhouse binary runs on distributions with old libc versions. If it fails, ask a maintainer for help	✅ success
Docker keeper image	The check to build and optionally push the mentioned image to docker hub	✅ success
Docker server image	The check to build and optionally push the mentioned image to docker hub	✅ success
Docs check	Builds and tests the documentation	✅ success
Fast test	Normally this is the first check that is ran for a PR. It builds ClickHouse and runs most of stateless functional tests, omitting some. If it fails, further checks are not started until it is fixed. Look at the report to see which tests fail, then reproduce the failure locally as described here	✅ success
Flaky tests	Checks if new added or modified tests are flaky by running them repeatedly, in parallel, with more randomization. Functional tests are run 100 times with address sanitizer, and additional randomization of thread scheduling. Integrational tests are run up to 10 times. If at least once a new test has failed, or was too long, this check will be red. We don't allow flaky tests, read the doc	✅ success
Install packages	Checks that the built packages are installable in a clear environment	✅ success
Integration tests	The integration tests report. In parenthesis the package type is given, and in square brackets are the optional part/total tests	✅ success
Mergeable Check	Checks if all other necessary checks are successful	✅ success
Performance Comparison	Measure changes in query performance. The performance test report is described in detail here. In square brackets are the optional part/total tests	✅ success
SQLancer	Fuzzing tests that detect logical bugs with SQLancer tool	✅ success
Sqllogic	Run clickhouse on the sqllogic test set against sqlite and checks that all statements are passed	✅ success
Stateful tests	Runs stateful functional tests for ClickHouse binaries built in various configurations -- release, debug, with sanitizers, etc	✅ success
Stateless tests	Runs stateless functional tests for ClickHouse binaries built in various configurations -- release, debug, with sanitizers, etc	✅ success
Stress test	Runs stateless functional tests concurrently from several clients to detect concurrency-related errors	✅ success
Style check	Runs a set of checks to keep the code style clean. If some of tests failed, see the related log from the report	✅ success
Unit tests	Runs the unit tests for different release types	✅ success
Upgrade check	Runs stress tests on server version from last release and then tries to upgrade it to the version from the PR. It checks if the new server can successfully startup without any errors, crashes or sanitizer asserts	✅ success

[nickitat]

it is not particularly relevant to this pr, but in general seems like there are quite a few places when any logging might be useful

[Stanislav-Vlasov]

I have a problem, I think, related to this change. I'm running the Clickhouse docker image version 24.2.2.71 (official build) on Ubuntu 22.04 (cgroup v2 enabled by default, no cgroup v1) with error in log. This error did not occur when I ran the previous version 24.1 of Clickhouse.

Application: Disabling cgroup memory observer because of an error during initialization: Code: 107. DB::Exception: Cannot find cgroups v1 or v2 current memory file. (FILE_DOESNT_EXIST), Stack trace (when copying this message, always include the lines below):

0. DB::Exception::Exception(DB::Exception::MessageMasked&&, int, bool) @ 0x000000000cf5af1b
1. DB::Exception::Exception<>(int, FormatStringHelperImpl<>) @ 0x00000000078cc583
2. DB::CgroupsMemoryUsageObserver::CgroupsMemoryUsageObserver(std::chrono::duration<long long, std::ratio<1l, 1l>>) @ 0x000000000d11da73
3. DB::Server::main(std::vector<String, std::allocator> const&) @ 0x000000000d0cfedd
4. Poco::Util::Application::run() @ 0x00000000158c9606
5. DB::Server::run() @ 0x000000000d0c0551
6. Poco::Util::ServerApplication::run(int, char**) @ 0x00000000158d2479
7. mainEntryClickHouseServer(int, char**) @ 0x000000000d0bd60a
8. main @ 0x00000000078c8e18
9. ? @ 0x00007f69f6e6d083
10. _start @ 0x00000000060c686e
    (version 24.2.2.71 (official build))

Additional information about my HOST system

1. uname -a
   Linux 6.2.0-37-generic 22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Thu Nov 2 18:01:13 UTC 2 x86_64 x86_64 x86_64 GNU/Linux

2. docker -v
   Docker version 24.0.2, build cb74dfc

Additional information from the inside of running container

1. cat /proc/1/cgroup
   0::/

2. ls /sys/fs/cgroup
   cgroup.controllers
   cgroup.events
   cgroup.freeze
   cgroup.kill
   cgroup.max.depth
   cgroup.max.descendants
   cgroup.pressure
   cgroup.procs
   cgroup.stat
   cgroup.subtree_control
   cgroup.threads
   cgroup.type
   cpu.idle
   cpu.max
   cpu.max.burst
   cpu.pressure
   cpuset.cpus
   cpuset.cpus.effective
   cpuset.cpus.partition
   cpuset.mems
   cpuset.mems.effective
   cpu.stat
   cpu.uclamp.max
   cpu.uclamp.min
   cpu.weight
   cpu.weight.nice
   hugetlb.1GB.current
   hugetlb.1GB.events
   hugetlb.1GB.events.local
   hugetlb.1GB.max
   hugetlb.1GB.numa_stat
   hugetlb.1GB.rsvd.current
   hugetlb.1GB.rsvd.max
   hugetlb.2MB.current
   hugetlb.2MB.events
   hugetlb.2MB.events.local
   hugetlb.2MB.max
   hugetlb.2MB.numa_stat
   hugetlb.2MB.rsvd.current
   hugetlb.2MB.rsvd.max
   io.bfq.weight
   io.max
   io.pressure
   io.prio.class
   io.stat
   io.weight
   memory.current
   memory.events
   memory.events.local
   memory.high
   memory.low
   memory.max
   memory.min
   memory.numa_stat
   memory.oom.group
   memory.peak
   memory.pressure
   memory.reclaim
   memory.stat
   memory.swap.current
   memory.swap.events
   memory.swap.high
   memory.swap.max
   memory.zswap.current
   memory.zswap.max
   misc.current
   misc.events
   misc.max
   pids.current
   pids.events
   pids.max
   pids.peak
   rdma.current
   rdma.max

[Stanislav-Vlasov]

As you can see, there are no nested cgroups in /sys/fs/cgroup

Due to problem of "Disabling cgroup memory observer", the available memory is incorrectly determined. Clickhouse believes that all HOST memory is available without respect of cgroup limit. This causes container memory overruns and triggers OOM killer.

[rschu1ze]

@Stanislav-Vlasov First, can you please open a GitHub issue for that? That would be a better place for discussion than the original PR.

As a workaround, you could set a sensible value for server setting max_server_memory_usage (docs). The error just means that ClickHouse will not react to memory limit changes done via cgroups.

Exception Cannot find cgroups v1 or v2 current memory file indicates that during database startup, neither a cgroups v1 nor v2 file was found. It would be interesting to debug this, perhaps you have a repro, e.g. the steps you use to invoke docker?

For completeness, I tested the latest official Docker image (v24.3, Ubuntu 22.04 LTS) and it worked for me, i.e. no errors.