Mime-Version: 1.0
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain; charset=UTF-8
Date: Sun, 10 Aug 2025 12:54:47 +0200
Message-Id: <DBYOXT4OXK6C.HZEDWO892EXN@mbosch.me>
Subject: Re: BUG #18991: NUMA test fails on machine with two NUMA nodes when
 building 18beta2
From: "Maximilian Bosch" <maximilian@mbosch.me>
To: <pgsql-bugs@lists.postgresql.org>
References: <18991-34c8f8afcd92c38b@postgresql.org>
In-Reply-To: <18991-34c8f8afcd92c38b@postgresql.org>
Archived-At: 
 <https://www.postgresql.org/message-id/DBYOXT4OXK6C.HZEDWO892EXN%40mbosch.me>
Precedence: bulk

Hi!

Quick follow-up from my side: I'd like to retract the bugreport, this is
entirely an issue on our side. The sandbox hides sysfs, so PostgreSQL
doesn't know about the NUMA topology at build-time, however the DB
running in the test still sees pages from several nodes which is
essentially what broke the test.

Apologies for the unnecessary bugreport, I know it can be frustrating
for upstreams to get reports that are essential "the distro messed it up".
I only do this when I'm absolutely certain that a distro bug can be
ruled out, this time I was wrong unfortunately.

Have a nice Sunday

Ma27

On Sat Jul 19, 2025 at 11:43 AM CEST, PG Bug reporting form wrote:
> The following bug has been logged on the website:
>
> Bug reference:      18991
> Logged by:          Maximilian Bosch
> Email address:      maximilian@mbosch.me
> PostgreSQL version: 18beta2
> Operating system:   NixOS 25.11 / unstable
> Description:       =20
>
> Hi!
>
> On NixOS master/unstable we provide betas of PostgreSQL to enable people
> such as extension devs to quickly identify problems with it.
>
> When building 18beta2 with `--with-libnuma` and linking against libnuma.s=
o
> from numactl 2.0.18, I get the following test failure while building the
> package on a machine with two NUMA nodes:
>
> ```
> diff -U3 /build/source/src/test/regress/expected/numa.out
> /build/source/src/test/regress/results/numa.out
> --- /build/source/src/test/regress/expected/numa.out    1970-01-01
> 00:00:01.000000000 +0000
> +++ /build/source/src/test/regress/results/numa.out     2025-07-19
> 08:44:02.793368816 +0000
> @@ -6,8 +6,4 @@
>  -- switch to superuser
>  \c -
>  SELECT COUNT(*) >=3D 0 AS ok FROM pg_shmem_allocations_numa;
> - ok
> -----
> - t
> -(1 row)
> -
> +ERROR:  invalid NUMA node id outside of allowed range [0, 0]: 1
> ```
>
> As mentioned above, the machine I used to build things has two NUMA nodes=
.
> `numactl --hardware` gives
>
> ```
> node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 2=
3
> 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 4=
8
> 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 7=
3
> 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 192 193
> 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 2=
12
> 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 2=
31
> 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 2=
50
> 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 2=
69
> 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287
> node 0 size: 773538 MB
> node 0 free: 603763 MB
> node 1 cpus: 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 =
112
> 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 1=
31
> 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 1=
50
> 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 1=
69
> 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 1=
88
> 189 190 191 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 3=
03
> 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 3=
22
> 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 3=
41
> 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 3=
60
> 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 3=
79
> 380 381 382 383
> node 1 size: 767886 MB
> node 1 free: 667997 MB
> node distances:
> node   0   1
>   0:  10  32
>   1:  32  10
> ```
>
> I'm aware that there was already discussion about this in
> https://www.postgresql.org/message-id/flat/E1u1tr8-003BbN-2E%40gemulon.po=
stgresql.org,
> but it seems like this was a different issue given that an error-code was
> given instead of an actual NUMA id? Also, all of the discussion including=
 a
> fix seem to have happened well before 18beta2, hence I'm filing a new
> report.
>
> Relevant downstream discussions:
> * https://github.com/NixOS/nixpkgs/pull/411958#issuecomment-3031660150
> * https://github.com/NixOS/nixpkgs/pull/426118
>
> Let me know if I can help with something, e.g. testing potential fixes.
>
> Best
> Ma27