Open lauri865 opened 9 months ago
Interestingly, the local CLI from my testing of PR 1003 is much faster than the current 0.12.3 on my laptop. Here's the numbers side by side:
Infisical CLI 0.12.3
[scott@samus infisical]$ time infisical secrets set FOO=BAR
┌─────────────┬──────────────┬────────────────┐
│ SECRET NAME │ SECRET VALUE │ STATUS │
├─────────────┼──────────────┼────────────────┤
│ FOO │ BAR │ SECRET CREATED │
└─────────────┴──────────────┴────────────────┘
real 0m1.331s
user 0m0.082s
sys 0m0.052s
[scott@samus infisical]$ time infisical secrets get FOO=BAR
┌─────────────┬──────────────┬─────────────┐
│ SECRET NAME │ SECRET VALUE │ SECRET TYPE │
├─────────────┼──────────────┼─────────────┤
│ FOO=BAR │ *not found* │ *not found* │
└─────────────┴──────────────┴─────────────┘
real 0m10.973s
user 0m0.081s
sys 0m0.054s
[scott@samus infisical]$ time infisical secrets delete --type shared FOO
secret name(s) [FOO] have been deleted from your project
real 0m0.707s
user 0m0.063s
sys 0m0.030s
Infisical compiled with go 1.20.7 for PR #1003:
[scott@samus infisical]$ time ./infisical-merge secrets set foo=bar
┌─────────────┬──────────────┬────────────────┐
│ SECRET NAME │ SECRET VALUE │ STATUS │
├─────────────┼──────────────┼────────────────┤
│ foo │ bar │ SECRET CREATED │
└─────────────┴──────────────┴────────────────┘
real 0m0.932s
user 0m0.070s
sys 0m0.050s
[scott@samus infisical]$ time ./infisical-merge secrets get foo
┌─────────────┬──────────────┬─────────────┐
│ SECRET NAME │ SECRET VALUE │ SECRET TYPE │
├─────────────┼──────────────┼─────────────┤
│ foo │ bar │ shared │
└─────────────┴──────────────┴─────────────┘
real 0m1.107s
user 0m0.265s
sys 0m0.122s
[scott@samus infisical]$ time ./infisical-merge secrets delete --type=shared foo
secret name(s) [foo] have been deleted from your project
real 0m0.517s
user 0m0.060s
sys 0m0.036s
That's a significant difference on secrets get . The current CLI is almost 11 seconds where my local compiled took 1.1s - against the same server/project from the same machine. The others commands were faster, too, but probably within a reasonable margin of error.
I think it's the newer go version (1.20 vs 1.19), since export has a similar improvement (10.961s vs 0.810s local-compile) and I didn't touch export.go in my PR.
What region are you making these cli calls from? @lauri865
In my case, Infisical is running on newer gen EPYC baremetal Kubernetes nodes that are roughly four miles away or about 20 minutes by bicycle in Southern California.
From Sweden, but it shouldn't matter. Network latency alone can't make up for this.
Same here. The run
command seems to take forever to run.
This definitely cannot be latency like @lauri865 suggested. My internet is >= 200Mbps at the moment
For some weird reason, this wasn't working until I did infisical login
and the run
command started finishing quickly. I'd let the token re-expire again and try to debug.
but definitely weird behaviour there
Describe the bug
Seeing considerably slow response times for infisical CLI.
E.g.
time infisical run --path=/test -- echo "done"
Can take around 8 seconds or more to run. For testing purposes, the path test contains only 1 secret. If there's referenced secret in the path, it becomes even slower by a few seconds.To Reproduce
Steps to reproduce the behavior:
Expected behavior
As a benchmark, the same equivalent command for
doppler run
takes 0.3-0.4s (20x faster). We would love to use an open source alternative, but the delta in response times impacts DX too much for us to pull the trigger at this time.Platform you are having the issue on:
OSX 13.5.1 (MBP with M2 Pro processor)
From SyncLinear.com | ENG-105