kemko/nomad
1
0
Fork 0
mirror of https://github.com/kemko/nomad.git synced 2026-01-03 17:05:43 +03:00
Code Issues Packages Projects Releases Wiki Activity

client/fingerprint/cpu: use fallback total compute value if cpu not detected

Browse Source 
Previously, Nomad would fail to startup if the CPU fingerprinter could
not detect the cpu total compute (i.e. cores * mhz). This is common on
some EC2 instance types (graviton class), where the env_aws fingerprinter
will override the detected CPU performance with a more accurate value
anyway.

Instead of crashing on startup, have Nomad use a low default for available
cpu performance of 1000 ticks (e.g. 1 core * 1 GHz). This enables Nomad
to get past the useless cpu fingerprinting on those EC2 instances. The
crashing error message is now a log statement suggesting the setting of
cpu_total_compute in client config.

Fixes #7989
This commit is contained in:
Seth Hoenig
2020-12-09 09:01:45 -06:00
parent e76add702d
commit da1235f35b
1 changed files with 14 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
client/fingerprint/cpu.go
View File

@@ -8,6 +8,14 @@ import (
"github.com/hashicorp/nomad/nomad/structs"
)
const (
// defaultCPUTicks is the default amount of CPU resources assumed to be
// available if the CPU performance data is unable to be detected. This is
// common on EC2 instances, where the env_aws fingerprinter will follow up,
// setting an accurate value.
defaultCPUTicks = 1000 // 1 core * 1 GHz
)
// CPUFingerprint is used to fingerprint the CPU
type CPUFingerprint struct {
StaticFingerprinter
@@ -64,12 +72,13 @@ func (f *CPUFingerprint) Fingerprint(req *FingerprintRequest, resp *FingerprintR
tt = cfg.CpuCompute
}
// Return an error if no cpu was detected or explicitly set as this
// node would be unable to receive any allocations.
// If we cannot detect the cpu total compute, fallback to a very low default
// value and log a message about configuring cpu_total_compute. This happens
// on Graviton instances where CPU information is unavailable. In that case,
// the env_aws fingerprinter updates the value with correct information.
if tt == 0 {
return fmt.Errorf("cannot detect cpu total compute. "+
"CPU compute must be set manually using the client config option %q",
"cpu_total_compute")
f.logger.Info("fallback to default cpu total compute, set client config option cpu_total_compute to override")
tt = defaultCPUTicks
}
resp.AddAttribute("cpu.totalcompute", fmt.Sprintf("%d", tt))