It does fail on OfBorg, so likely it would fail when I build it too.

I'll see if I can update it in the coming days, I recall it being somewhat painful to update the last time as well, but I do have some beefy machines that can at least build it, so should be possible

We should probably switch clickhouse to using UNBUNDLED mode instead so we don't have to rebuild e.g. LLVM as part of the clickhouse build...

I hit

../utils/db-generator/query_db_generator.cpp: In function 'int main(int, const char**)':
../utils/db-generator/query_db_generator.cpp:1295:20: error: ignoring return value of 'FILE* freopen(const char*, const char*, FILE*)', declared with attribute warn_unused_result [-Werror=unused-result]
 1295 |             freopen(vm["input"].as<std::string>().c_str(), "r", stdin);
      |             ~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../utils/db-generator/query_db_generator.cpp:1297:20: error: ignoring return value of 'FILE* freopen(const char*, const char*, FILE*)', declared with attribute warn_unused_result [-Werror=unused-result]
 1297 |             freopen(vm["output"].as<std::string>().c_str(), "w", stdout);
      |             ~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
cc1plus: all warnings being treated as errors

while building this.

#105286 makes Clickhouse use the system LLVM instead of compiling its own, which at least seemed to make it evaluate on ofBorg.

There is a patch for the build issue in: ClickHouse/ClickHouse#16859

Until there is a new upstream version, we could apply that patch ourselves

With

diff --git a/pkgs/servers/clickhouse/default.nix b/pkgs/servers/clickhouse/default.nix
index b519d1b4315..2cad2150153 100644
--- a/pkgs/servers/clickhouse/default.nix
+++ b/pkgs/servers/clickhouse/default.nix
@@ -1,4 +1,4 @@
-{ stdenv, fetchFromGitHub, cmake, libtool, lldClang, ninja
+{ stdenv, fetchFromGitHub, fetchpatch, cmake, libtool, lldClang, ninja
 , boost, brotli, capnproto, cctz, clang-unwrapped, double-conversion
 , icu, jemalloc, libcpuid, libxml2, lld, llvm, lz4, libmysqlclient, openssl, perl
 , poco, protobuf, python3, rapidjson, re2, rdkafka, readline, sparsehash, unixODBC
@@ -17,6 +17,13 @@ stdenv.mkDerivation rec {
     sha256 = "0c87k0xqwj9sc3xy2f3ngfszgjiz4rzd787bdg6fxp94w1adjhny";
   };
 
+  patches = [
+    (fetchpatch {
+      url = "https://github.com/ClickHouse/ClickHouse/commit/e31753b4db7aa0a72a85757dc11fc403962e30db.patch";
+      sha256 = "12ax02dh9y9k8smkj6v50yfr46iprscbrvd4bb9vfbx8xqgw7grb";
+    })
+  ];
+
   nativeBuildInputs = [ cmake libtool lldClang.bintools ninja ];
   buildInputs = [
     boost brotli capnproto cctz clang-unwrapped double-conversion

This builds for me:

[lukegb@whitby:~/nixpkgs-clickhouse]$ ./result/bin/clickhouse server --version
ClickHouse server version 20.11.4.13.

[lukegb@whitby:~/nixpkgs-clickhouse]$ ./result/bin/clickhouse client --version
ClickHouse client version 20.11.4.13.

@schneefux You should be able to git fetch https://github.com/NixOS/nixpkgs.git eb960b541dfc9a18a4bb4fd7b8a09757e97d14fe && git cherry-pick eb960b541dfc9a18a4bb4fd7b8a09757e97d14fe.

You'll also want to rebase on master, since it now has my change to use system LLVM instead of the bundled one, which should make it build (much) faster. I'll rebuild once you've pushed up a new branch :)

Result of nixpkgs-review pr 105224 run on x86_64-linux 1

It still does not build within an hour on my server, so I'm afraid I have to give up and hand it over to @RaghavSood 🙈
Thank you for your help!

Result of nixpkgs-review pr 105224 run on x86_64-linux 1

  
  CMake Error at cmake/find/llvm.cmake:50 (message):
    Unsupported LLVM configuration, cannot enable LLVM
  Call Stack (most recent call first):
    CMakeLists.txt:440 (include)

hrm. Maybe I broke it.

Clickhouse now needs LLVM10: you'll need to update pkgs/top-level/all-packages.nix to change the clickhouse customization to use llvmPackages10 instead of llvmPackages9, but that should be sufficient to get it to build - see the OfBorg failure: https://logs.nix.ci/?key=nixos/nixpkgs.105224&attempt_id=8b71ccee-d166-4ed4-a69c-cdb990c038f1

Continuing in #105968