build.sh hangs on a large machine

[omajid]

Describe the Bug

The CentOS Stream 10 builders have a beefy machine with the specs described below. On this machine, running baremetal (not virtualized), the .NET VMR fails to build. The build just hangs.

CPU info:
Architecture:        x86_64
CPU op-mode(s):      32-bit, 64-bit
Byte Order:          Little Endian
CPU(s):              48
On-line CPU(s) list: 0-47
Thread(s) per core:  2
Core(s) per socket:  12
Socket(s):           2
NUMA node(s):        2
Vendor ID:           GenuineIntel
CPU family:          6
Model:               106
Model name:          Intel(R) Xeon(R) Silver 4310 CPU @ 2.10GHz
Stepping:            6
CPU MHz:             3300.000
CPU max MHz:         3300.0000
CPU min MHz:         800.0000
BogoMIPS:            4200.00
Virtualization:      VT-x
L1d cache:           48K
L1i cache:           32K
L2 cache:            1280K
L3 cache:            18432K
NUMA node0 CPU(s):   0,2,4,6,8,10,12,14,16,18,20,22,24,26,28,30,32,34,36,38,40,42,44,46
NUMA node1 CPU(s):   1,3,5,7,9,11,13,15,17,19,21,23,25,27,29,31,33,35,37,39,41,43,45,47
Flags:               fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 invpcid_single intel_ppin ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local split_lock_detect wbnoinvd dtherm ida arat pln pts avx512vbmi umip pku ospke avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq la57 rdpid fsrm md_clear pconfig flush_l1d arch_capabilities


Memory:
              total        used        free      shared  buff/cache   available
Mem:      263557768    18390244   126431532     3276688   118735992   240039372
Swap:       2097148      816068     1281080

This does not happen on any other environment I have seen. Even other CentOS Stream 10 builders work just fine.

Could this be a scaling issue where the cpu:memory ratio is off?

[tmds]

Could this be a scaling issue where the cpu:memory ratio is off?

Or perhaps the multiple numa nodes are causing some issue.

The hang occurs while building with the Microsoft SDK. It's not specific to a source-build SDK.

cc @jkotas

[jkotas]

Could this be a scaling issue where the cpu:memory ratio is off?

Scaling issue would manifest as build running slower than expected. If you see a hang, it is likely a bug (race condition) that got exposed by running on a large machine.

Could you please attach debugger to the hung process(es) and find where they are stuck (collect stacktraces of all threads)?

[omajid]

Could you please attach debugger to the hung process(es) and find where they are stuck (collect stacktraces of all threads)?

Sorry, I can't. The one machine where this is reproducible is part of a locked down build farm that does not have interactive access enabled. I haven't been able to reproduce this on any other machine where I can do any interactive debugging.

[jkotas]

Are you able to download logs and crash dumps form these machines?

• You can try to use dotnet-stack on all processes in the hung state
  and/or
• You can enable https://github.com/dotnet/runtime/blob/c2f55e0eca1c8a7dfe2cd6899c95d24ab235ad1c/src/coreclr/pal/src/exception/signal.cpp#L850, and send all dotnet processes SIGTERM signal that should produce crash dumps