few weeks ago i’ve done a general firmware upgrade of 2 identical Dell Poweredge r620 servers. a while later i’ve noticed that one of them had really sluggish performances. i’ve noticed that /proc/cpuinfo has shown CPU frequencies of hundreds of MHz rather than thousands.
processor : 15 vendor_id : GenuineIntel cpu family : 6 model : 45 model name : Intel(R) Xeon(R) CPU E5-2643 0 @ 3.30GHz stepping : 7 microcode : 0x710 cpu MHz : 163.751 cache size : 10240 KB physical id : 1 siblings : 8 core id : 3 cpu cores : 4 apicid : 39 initial apicid : 39 fpu : yes fpu_exception : yes cpuid level : 13 wp : yes 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 arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf eagerfpu pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic popcnt tsc_deadline_timer aes xsave avx lahf_lm tpr_shadow vnmi flexpriority ept vpid xsaveopt dtherm ida arat pln pts bugs : bogomips : 6634.62 clflush size : 64 cache_alignment : 64 address sizes : 46 bits physical, 48 bits virtual power management:
cat /proc/cpuinfo |grep MHz gave me 16 rows of values between 150 and 200MHz, while – on the other identical machine with the same kernel 4.9 from Debian Stretch – values would range from 1199.835MHz to 3494.262MHz.
i’ve googled and stumbled on:
i’ve used openssl speed as a simple, non-scientific way of comparing the computing power between the problematic and good servers. on the good one i’d get Doing md4 for 3s on 16 size blocks: 12610373 md4’s in 3.00s, on the underperforming – values ranging between 3000000 and 7000000.
what i’ve tried without much luck:
- boot from an ubuntu live dvd – both /proc/cpuinfo and openssl speed were as bad
- boot from OpenIndiana Hipster 2017.10 live dvd – the same
- install under debian intel-microcode from the stretch-backports – the same
- power off and power on the server few times – no luck
- boot linux with the kernel switch intel_pstate=disable – no joy
- checked bios, firmware versions, settings between the two identical servers
- reset bios settings to defaults, try changing the power profile to performance again
at the end i’ve downgraded BIOS version from 2.5.4 to 2.4.3 – the problem was gone! then i went one up to 2.5.2 – all was still fine… and then to 2.5.4 and it’s still fine. go figure.