Invalid peer certificate on ztunnel, when using subordinate AWS private CA #1061
Comments
guilhermef
changed the title
|
Working on a repro now. Will share an update soon! |
|
Can you provide more info on how you are using an AWS Private CA ? MeshConfig, what are the expectations, etc. ztunnel by default is using Istiod CA. |
|
Hi @costinm, the istio root CA is a subordinate of the AWS private root CA, we use cert-manager AWS PCA issuer to create the certificate for istio. |
|
Ah, so were you using the plugin CA feature of Istio? |
|
Exactly, @keithmattix; we use Istio to connect multiple clusters hosted in AWS. |
|
Usually it means either that ztunnel is not sending the full chain or that the top root is not properly distributed. Not sure if we have some debug level that dumps the chain - perhaps a curl -vvv or openssl client connecting to ztunnel can show the chain that is used ( should have 3 certs instead of 2). If that looks right - probably the root cert is the problem. We should add some debug, istiod does show the root at startup. Normally ztunnel should get all the roots from mesh config too. |
|
@costinm, that's all I got from ztunnel 15008 port. |
|
@costinm, here are the debug logs from a ztunnel pod. |
|
Deja vue - but we'll need to reproduce and add more info ( and fix the bug
). Are you using certmanager ? Can you include the configs you used to
setup the certs - install options, crs, etc ?
There are unfortunately many ways to setup the roots and CA - would help to
reproduce yours (or cluse enough).
…On Fri, May 17, 2024, 16:00 Guilherme Souza ***@***.***> wrote:
@costinm <https://github.com/costinm>, here are the debug logs from a
ztunnel pod.
2024-05-18 00:55:41.082 | 2024-05-17T22:55:41.082082Z debug h2::codec::framed_write:xds{id=1}:Connection{peer=Client} send frame=WindowUpdate { stream_id: StreamId(0), size_increment: 5177345 } |
-- | -- | --
| | 2024-05-18 00:55:41.082 | 2024-05-17T22:55:41.082056Z debug hyper_util::client::legacy::pool:xds{id=1} pooling idle connection for ("https", istiod.istio-system.svc:15012) |
| | 2024-05-18 00:55:41.082 | 2024-05-17T22:55:41.082013Z debug h2::codec::framed_write:xds{id=1} send frame=Settings { flags: (0x0), enable_push: 0, initial_window_size: 2097152, max_frame_size: 16384, max_header_list_size: 16384 } |
| | 2024-05-18 00:55:41.082 | 2024-05-17T22:55:41.081995Z debug h2::client:xds{id=1} client connection bound |
| | 2024-05-18 00:55:41.082 | 2024-05-17T22:55:41.081983Z debug h2::client:xds{id=1} binding client connection |
| | 2024-05-18 00:55:41.081 | 2024-05-17T22:55:41.081687Z debug rustls::client::common:xds{id=1} Client auth requested but no cert/sigscheme available |
| | 2024-05-18 00:55:41.081 | 2024-05-17T22:55:41.081670Z debug rustls::client::tls13:xds{id=1} Got CertificateRequest CertificateRequestPayloadTls13 { context: , extensions: [Unknown(UnknownExtension { typ: StatusRequest, payload: }), Unknown(UnknownExtension { typ: SCT, payload: }), SignatureAlgorithms([RSA_PSS_SHA256, ECDSA_NISTP256_SHA256, ED25519, RSA_PSS_SHA384, RSA_PSS_SHA512, RSA_PKCS1_SHA256, RSA_PKCS1_SHA384, RSA_PKCS1_SHA512, ECDSA_NISTP384_SHA384, ECDSA_NISTP521_SHA512, RSA_PKCS1_SHA1, ECDSA_SHA1_Legacy]), AuthorityNames([DistinguishedName(30183116301406035504030c0d53756d757020526f6f74204341)])] } |
| | 2024-05-18 00:55:41.081 | 2024-05-17T22:55:41.081657Z debug rustls::client::hs:xds{id=1} ALPN protocol is Some(b"h2") |
| | 2024-05-18 00:55:41.081 | 2024-05-17T22:55:41.081647Z debug rustls::client::tls13:xds{id=1} TLS1.3 encrypted extensions: [Protocols([ProtocolName(6832)])] |
| | 2024-05-18 00:55:41.081 | 2024-05-17T22:55:41.081541Z debug rustls::client::tls13:xds{id=1} Not resuming |
| | 2024-05-18 00:55:41.081 | 2024-05-17T22:55:41.081520Z debug rustls::client::hs:xds{id=1} Using ciphersuite TLS13_AES_128_GCM_SHA256 |
| | 2024-05-18 00:55:41.078 | 2024-05-17T22:55:41.078873Z debug rustls::client::hs:xds{id=1} Not resuming any session |
| | 2024-05-18 00:55:41.078 | 2024-05-17T22:55:41.078727Z debug rustls::client::hs:xds{id=1} No cached session for DnsName("istiod.istio-system.svc") |
| | 2024-05-18 00:55:41.078 | 2024-05-17T22:55:41.078700Z debug hyper_util::client::legacy::connect::http:xds{id=1} connected to 172.20.100.179:15012 |
| | 2024-05-18 00:55:41.077 | 2024-05-17T22:55:41.077901Z debug hyper_util::client::legacy::connect::http:xds{id=1} connecting to 172.20.100.179:15012 |
| | 2024-05-18 00:55:41.076 | 2024-05-17T22:55:41.076231Z debug hyper_util::client::legacy::connect::dns resolving host="istiod.istio-system.svc" |
| | 2024-05-18 00:55:41.076 | 2024-05-17T22:55:41.076079Z debug rustls::webpki::anchors:xds{id=1} add_parsable_certificates processed 1 valid and 0 invalid certs |
| | 2024-05-18 00:55:41.075 | 2024-05-17T22:55:41.075931Z info readiness Task 'proxy' complete (1.174111ms), still awaiting 2 tasks |
| | 2024-05-18 00:55:41.075 | 2024-05-17T22:55:41.075924Z info hyper_util listener established address=[::]:15020 component="stats" |
| | 2024-05-18 00:55:41.075 | 2024-05-17T22:55:41.075898Z info hyper_util listener established address=127.0.0.1:15000 component="admin" |
| | 2024-05-18 00:55:41.075 | 2024-05-17T22:55:41.075829Z info hyper_util listener established address=[::]:15021 component="readiness" |
| | 2024-05-18 00:55:41.075 | 2024-05-17T22:55:41.075815Z info app in-pod mode enabled |
| | 2024-05-18 00:55:41.075 | 2024-05-17T22:55:41.075530Z debug rustls::webpki::anchors add_parsable_certificates processed 1 valid and 0 invalid certs |
| | 2024-05-18 00:55:41.075 | |
| | 2024-05-18 00:55:41.075 | inpodMark: 1337 |
| | 2024-05-18 00:55:41.075 | inpodPortReuse: true |
| | 2024-05-18 00:55:41.075 | inpodUds: /var/run/ztunnel/ztunnel.sock |
| | 2024-05-18 00:55:41.075 | inpodEnabled: true |
| | 2024-05-18 00:55:41.075 | shuffle_dns_servers: false |
| | 2024-05-18 00:55:41.075 | authentic_data: false |
| | 2024-05-18 00:55:41.075 | recursion_desired: true |
| | 2024-05-18 00:55:41.075 | server_ordering_strategy: QueryStatistics |
| | 2024-05-18 00:55:41.075 | try_tcp_on_error: false |
| | 2024-05-18 00:55:41.075 | preserve_intermediates: true |
| | 2024-05-18 00:55:41.075 | num_concurrent_reqs: 2 |
| | 2024-05-18 00:55:41.075 | negative_max_ttl: null |
| | 2024-05-18 00:55:41.075 | positive_max_ttl: null |
| | 2024-05-18 00:55:41.075 | negative_min_ttl: null |
| | 2024-05-18 00:55:41.075 | positive_min_ttl: null |
| | 2024-05-18 00:55:41.075 | use_hosts_file: true |
| | 2024-05-18 00:55:41.075 | cache_size: 32 |
| | 2024-05-18 00:55:41.075 | ip_strategy: Ipv4thenIpv6 |
| | 2024-05-18 00:55:41.075 | validate: false |
| | 2024-05-18 00:55:41.075 | edns0: false |
| | 2024-05-18 00:55:41.075 | check_names: true |
| | 2024-05-18 00:55:41.075 | rotate: false |
| | 2024-05-18 00:55:41.075 | attempts: 2 |
| | 2024-05-18 00:55:41.075 | nanos: 0 |
| | 2024-05-18 00:55:41.075 | secs: 5 |
| | 2024-05-18 00:55:41.075 | timeout: |
| | 2024-05-18 00:55:41.075 | ndots: 5 |
| | 2024-05-18 00:55:41.075 | dnsResolverOpts: |
| | 2024-05-18 00:55:41.075 | bind_addr: null |
| | 2024-05-18 00:55:41.075 | trust_negative_responses: false |
| | 2024-05-18 00:55:41.075 | tls_dns_name: null |
| | 2024-05-18 00:55:41.075 | protocol: tcp |
| | 2024-05-18 00:55:41.075 | - socket_addr: 172.20.0.10:53 |
| | 2024-05-18 00:55:41.075 | bind_addr: null |
| | 2024-05-18 00:55:41.075 | trust_negative_responses: false |
| | 2024-05-18 00:55:41.075 | tls_dns_name: null |
| | 2024-05-18 00:55:41.075 | protocol: udp |
| | 2024-05-18 00:55:41.075 | - socket_addr: 172.20.0.10:53 |
| | 2024-05-18 00:55:41.075 | name_servers: |
| | 2024-05-18 00:55:41.075 | - eu-west-1.compute.internal |
| | 2024-05-18 00:55:41.075 | - cluster.local |
| | 2024-05-18 00:55:41.075 | - svc.cluster.local |
| | 2024-05-18 00:55:41.075 | - istio-system.svc.cluster.local |
| | 2024-05-18 00:55:41.075 | search: |
| | 2024-05-18 00:55:41.075 | domain: null |
| | 2024-05-18 00:55:41.075 | dnsResolverCfg: |
| | 2024-05-18 00:55:41.075 | proxyArgs: proxy ztunnel |
| | 2024-05-18 00:55:41.075 | enableOriginalSource: null |
| | 2024-05-18 00:55:41.075 | numWorkerThreads: 2 |
| | 2024-05-18 00:55:41.075 | CLUSTER_ID: fleet-sandbox-eu-west-1 |
| | 2024-05-18 00:55:41.075 | ISTIO_VERSION: 1.22.0 |
| | 2024-05-18 00:55:41.075 | DNS_PROXY_ADDR: 127.0.0.1:15053 |
| | 2024-05-18 00:55:41.075 | proxyMetadata: |
| | 2024-05-18 00:55:41.075 | nanos: 0 |
| | 2024-05-18 00:55:41.075 | secs: 5 |
| | 2024-05-18 00:55:41.075 | selfTerminationDeadline: |
| | 2024-05-18 00:55:41.075 | fakeCa: false |
| | 2024-05-18 00:55:41.075 | xdsOnDemand: false |
| | 2024-05-18 00:55:41.075 | nanos: 0 |
| | 2024-05-18 00:55:41.075 | secs: 86400 |
| | 2024-05-18 00:55:41.075 | secretTtl: |
| | 2024-05-18 00:55:41.075 | xdsRootCert: !File ./var/run/secrets/istio/root-cert.pem |
| | 2024-05-18 00:55:41.075 | xdsAddress: https://istiod.istio-system.svc:15012 |
| | 2024-05-18 00:55:41.075 | caRootCert: !File ./var/run/secrets/istio/root-cert.pem |
| | 2024-05-18 00:55:41.075 | caAddress: https://istiod.istio-system.svc:15012 |
| | 2024-05-18 00:55:41.075 | clusterDomain: cluster.local |
| | 2024-05-18 00:55:41.075 | clusterId: fleet-sandbox-eu-west-1 |
| | 2024-05-18 00:55:41.075 | localIp: 10.61.86.165 |
| | 2024-05-18 00:55:41.075 | proxyMode: Shared |
| | 2024-05-18 00:55:41.075 | localNode: ip-10-61-94-2.eu-west-1.compute.internal |
| | 2024-05-18 00:55:41.075 | network: '' |
| | 2024-05-18 00:55:41.075 | dnsProxyAddr: 127.0.0.1:15053 |
| | 2024-05-18 00:55:41.075 | outboundAddr: '[::]:15001' |
| | 2024-05-18 00:55:41.075 | inboundPlaintextAddr: '[::]:15006' |
| | 2024-05-18 00:55:41.075 | inboundAddr: '[::]:15008' |
| | 2024-05-18 00:55:41.075 | readinessAddr: '[::]:15021' |
| | 2024-05-18 00:55:41.075 | statsAddr: '[::]:15020' |
| | 2024-05-18 00:55:41.075 | adminAddr: 127.0.0.1:15000 |
| | 2024-05-18 00:55:41.075 | socks5Addr: null |
| | 2024-05-18 00:55:41.075 | nanos: 0 |
| | 2024-05-18 00:55:41.075 | secs: 300 |
| | 2024-05-18 00:55:41.075 | poolUnusedReleaseTimeout: |
| | 2024-05-18 00:55:41.075 | poolMaxStreamsPerConn: 100 |
| | 2024-05-18 00:55:41.075 | frameSize: 1048576 |
| | 2024-05-18 00:55:41.075 | connectionWindowSize: 4194304 |
| | 2024-05-18 00:55:41.075 | windowSize: 4194304 |
| | 2024-05-18 00:55:41.075 | dnsProxy: false |
| | 2024-05-18 00:55:41.075 | 2024-05-17T22:55:41.075306Z info ztunnel running with config: proxy: true |
| | 2024-05-18 00:55:41.075 | 2024-05-17T22:55:41.075092Z info ztunnel version: version.BuildInfo{Version:"909bf991d01edc4db51265bc633acfe303555ef5", GitRevision:"909bf991d01edc4db51265bc633acfe303555ef5", RustVersion:"1.77.2", BuildProfile:"release", BuildStatus:"Clean", GitTag:"1.22.0-beta.1-6-g909bf99", IstioVersion:"unknown"}
—
Reply to this email directly, view it on GitHub
<#1061 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAAUR2XNQEKQAALWLKC427DZC2DZRAVCNFSM6AAAAABHZ3NTZWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCMJYGQ3TANBTGI>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
|
Yes, @costinm, we're using cert-manager. apiVersion: cert-manager.io/v1
kind: Certificate
metadata:
name: istio-ca
namespace: istio-system
spec:
commonName: istio-ca
duration: 720h0m0s
isCA: true
issuerRef:
group: awspca.cert-manager.io
kind: AWSPCAIssuer
name: aws-pca-istio-issuer
renewBefore: 168h0m0s
secretName: cacerts
subject:
organizations:
- cluster.local
- cert-manager
status:
conditions:
- lastTransitionTime: "2024-04-11T12:16:39Z"
message: Certificate is up to date and has not expired
observedGeneration: 1
reason: Ready
status: "True"
type: Ready
notAfter: "2024-06-03T12:16:36Z"
notBefore: "2024-05-04T11:16:36Z"
renewalTime: "2024-05-27T12:16:36Z"
revision: 2apiVersion: awspca.cert-manager.io/v1beta1
kind: AWSPCAIssuer
metadata:
name: aws-pca-istio-issuer
namespace: istio-system
spec:
arn: arn:aws:acm-pca:eu-west-1:<redacted>:certificate-authority/<redacted>
region: eu-west-1
status:
conditions:
- lastTransitionTime: "2024-04-11T12:16:36Z"
message: Issuer verified
reason: Verified
status: "True"
type: ReadyapiVersion: cert-manager.io/v1
kind: Issuer
metadata:
name: istio-ca
namespace: istio-system
spec:
ca:
secretName: cacerts
status:
conditions:
- lastTransitionTime: "2024-04-11T12:16:41Z"
message: Signing CA verified
observedGeneration: 1
reason: KeyPairVerified
status: "True"
type: Ready
|
|
Ok, took a bit of a look at the code - there are some problems with cacerts if tls.key is used and multiple intermediaries are used ( handled correctly on the old code path ) - but for this case it should work (the 'intermediary' chain is created by only 2 certs, since CertManager doesn't have a way to pass a longer chain AFAIK. Can you check the cert created by AWS and make sure it's just a root + a cert ( not a longer chain ) ? The code is pretty convoluted - you may want to create an 'old style' cacerts manually (using cert-chain.pem has the entire chain starting with the CA cert associated with the key and all the way to the real root ). Again - this is in the case the chain has more than 2 elements, with 2 I think the code should work the same in both cases. I looked a bit at the code in ztunnel - I'm not very good at rust, but it looks like caclient.rs is handling the cert response expecting a list of PEM files and does the right thing - can't repro the AWS case, but if you see more than 2 certs in the chains it's very likely to be the problem, and you can confirm by using the old-style cert-chain.pem If you only see 2 certs in the AWS chain ( the top root and the intermediary ) - we'll need more debugging. Unfortunately the code path loading from Secret doesn't work the same, so harder to reproduce in a debugger. |
|
@costinm, that's how the secret is created by cert-manager when using AWS PCA: ca.crt: |
-----BEGIN CERTIFICATE-----
<cert-data>
-----END CERTIFICATE-----
tls.crt: |
-----BEGIN CERTIFICATE-----
<cert-data>
-----END CERTIFICATE-----
-----BEGIN CERTIFICATE-----
<cert-data>
-----END CERTIFICATE-----
-----BEGIN CERTIFICATE-----
<cert-data>
-----END CERTIFICATE-----
tls.key: |
-----BEGIN RSA PRIVATE KEY-----
<cert-data>
-----END RSA PRIVATE KEY-----We do have 3 certs in |
|
cert-manager does offer a way to create a combined .pem file, but it's in alpha: |
|
@costinm, I just tested with a root CA, instead of the subordinate CA that we were using previously, and it works. |
guilhermef
changed the title
To confirm, is this the data stored in the |
|
I was able to create deeper intermediary with certmanager and it seems to also work. Need to double check the config. Are you sure the top level root ( last one in the chain ) is set in the secret and replicated ? |
When you inspect your |
|
Hi @jaellio and @costinm, yes, the YAML I shared previously is the contents of I just tested another setup, with ca.crt: |
-----BEGIN CERTIFICATE-----
<Root CA>
-----END CERTIFICATE-----
tls.crt: |
-----BEGIN CERTIFICATE-----
<cert-data>
-----END CERTIFICATE-----
-----BEGIN CERTIFICATE-----
<cert-data>
-----END CERTIFICATE-----
tls.key: |
-----BEGIN RSA PRIVATE KEY-----
<cert-data>
-----END RSA PRIVATE KEY-----The On this second case, the result was the same, we also got an |
|
I think chained certs are working fine... Cany you check: and make sure it's the same - and the Main CA root you use ? |
|
To step back - unknown Issuer means something in the chain is not known or trusted. Original guess was that the chain is bad (missing certs), but it could be that the roots are wrong. Istio does send the full chain including the top root (never understood why...) - but the actual check is against the istio-ca-root-cert which is a list of roots - and must include the root that signed the chain. |
|
Hi @costinm, I just checked the |
|
I 'm adding more debug info on what CAs are loaded and fixing the inconsistencies - but not sure I can help more with your case, for the intermediary I created with CertManager it seems fine. Can you repeat the 'openssl s_client -connect 10.61.73.159:15008' command - but make sure you are connecting to a pod IP ( not ztunnel IP ), and from a VM or pod that is not ambient enabled ? Checked with John - ztunnel IP doesn't know what cert to return, it relies on detecting the intended pod and serving its certificate. |
|
@costinm, now I was able to get the certificate data: Connecting to 10.61.72.122
CONNECTED(00000003)
Can't use SSL_get_servername
depth=1 O=cert-manager + O=cluster.local, CN=istio-ca
verify error:num=20:unable to get local issuer certificate
verify return:1
depth=0
verify return:1
---
Certificate chain
0 s:
i:O=cert-manager + O=cluster.local, CN=istio-ca
a:PKEY: id-ecPublicKey, 256 (bit); sigalg: RSA-SHA256
v:NotBefore: May 30 20:56:54 2024 GMT; NotAfter: May 31 20:58:54 2024 GMT
1 s:O=cert-manager + O=cluster.local, CN=istio-ca
i:O=Company, OU=Platform, CN=Company Istio sandbox CA
a:PKEY: rsaEncryption, 2048 (bit); sigalg: RSA-SHA256
v:NotBefore: May 24 04:18:20 2024 GMT; NotAfter: Jun 23 05:18:20 2024 GMT
---
Server certificate
-----BEGIN CERTIFICATE-----
<CERT_DATA>
-----END CERTIFICATE-----
subject=
issuer=O=cert-manager + O=cluster.local, CN=istio-ca
---
Acceptable client certificate CA names
CN=Company Root CA
Requested Signature Algorithms: ECDSA+SHA384:ECDSA+SHA256:Ed25519:RSA-PSS+SHA512:RSA-PSS+SHA384:RSA-PSS+SHA256:RSA+SHA512:RSA+SHA384:RSA+SHA256
Shared Requested Signature Algorithms: ECDSA+SHA384:ECDSA+SHA256:Ed25519:RSA-PSS+SHA512:RSA-PSS+SHA384:RSA-PSS+SHA256:RSA+SHA512:RSA+SHA384:RSA+SHA256
Peer signing digest: SHA256
Peer signature type: ECDSA
Server Temp Key: X25519, 253 bits
---
SSL handshake has read 2209 bytes and written 409 bytes
Verification error: unable to get local issuer certificate
---
New, TLSv1.3, Cipher is TLS_AES_256_GCM_SHA384
Server public key is 256 bit
This TLS version forbids renegotiation.
No ALPN negotiated
Early data was not sent
Verify return code: 20 (unable to get local issuer certificate)
---
287B1D54FD7E0000:error:0A00045C:SSL routines:ssl3_read_bytes:tlsv13 alert certificate required:ssl/record/rec_layer_s3.c:907:SSL alert number 116 |
|
We're getting somewhere. Looks like it's too short - missing a bunch of certs. Will take a look again at the code - missing the full chain was my first guess, but when I tested it seemed to work. |
|
@costinm based on the chain that I got from the istio secret, it's missing the Istio CA. On the Istio secret, it contains the three certificates. |
|
Sorry - long thread maybe I missed it, there was a question on which secret you use - istio-ca-secret or cacerts ? Different code paths ... |
|
@costinm, we're using |
|
Currently running into the same error, and what appears to be the same issue (thanks @howardjohn for linking me here). My setup consists of:
The The The following is output of Here is the output of From my reading of the Thanks for investigating this! I can try to create the Happy to provide debug logs or additional details. |
I'm testing ztunnel on 1.22 release.
I have 2 book example apps deployed, one with sidecar and one with ambient.
The sidecar one is working, but with the ambient one I get
io error: invalid peer certificate: UnknownIssuer.We're using an AWS Private CA.
ztunnel logs:
istioctl x ztunnel-config certificates ztunnel-9tkd9.istio-systemistioctl x ztunnel-config certificates ztunnel-dxgrd.istio-systemIf I run ambient locally, with the certificate created by Istio, there's no intermediate certificate.
Update:
There's an issue with zTunnel when using AWS PCA and cert-manager, with a subordinate CA.
The text was updated successfully, but these errors were encountered: