Is there a straight forward way to profile a refresh This st

Question

Is there a straight forward way to profile a refresh This stack has like 20 some odd vpcs and some of them are peered but I can t help but feel it s refreshing the same resources several times It s upwards of 30 minutes to run a refresh I ve done equally complex resources layouts in other IaC languages and even 10 minutes is long

white-balloon-205 · Answer

You can grab a trace with the command at <https github com pulumi pulumi issues 3257 issuecomment 534629119> the same should work for a `refresh` Then share the trace here or on a GitHub issue 30 minutes does indeed sound like a very long time for any refresh

worried-city-86458 · Answer

I m finding my patience is being test as well so I m really looking forward to any perf improvements

white-balloon-205 · Answer

< worried city 86458> On what operations exactly refresh up other And I forget have you shared a trace yet

worried-city-86458 · Answer

Up operations using Helm charts in particular

worried-city-86458 · Answer

If you can make the Istio chart fast that should cover most bases stuck out tongue

white-balloon-205 · Answer

Got it if you haven t yet shared a trace it would be great to see one that is typical of where things feel particularly slow just so we can verify Though for large helm charts I m guessing this is related to the issues discussed at the bottom of <https github com pulumi pulumi issues 3257> I m hoping we ll be able to tackle both of those very soon

worried-city-86458 · Answer

I m not using it yet but heading that way

acceptable-army-69872 · Answer

I m running again with trace The last refresh bombed out which could have been my keys expiring

acceptable-army-69872 · Answer

``` Resources + 1260 to create ~ 130 to update 179 to delete + 290 to replace 1859 changes 256 unchanged error failed to complete update 403 Forbidden ```

white-balloon-205 · Answer

BTW encourage anyone seeing perf issues to upvote 3257 and share traces or details of their use cases we re actively working on improvements here so the more data we have to guide this work the better

worried-city-86458 · Answer

I m watching for any k8s changes and happy to try anything in alpha

acceptable-army-69872 · Answer

< white balloon 205> the trace is 305 megs `305M up trace` When I hit the URL I get two request auth 400ish ms and the this pulumi trace and when I try to load the pulumi trace my browser crashes

gentle-diamond-70147 · Answer

< acceptable army 69872> can you send us the file s3 url

acceptable-army-69872 · Answer

It compressed down to 29 megs

gorgeous-egg-16927 · Answer

Released version 1 2 1 of the k8s provider that significantly improves performance of preview on stacks with a lot of k8s resources

gorgeous-egg-16927 · Answer

We re working on several related PRs that should help out a lot