-
Notifications
You must be signed in to change notification settings - Fork 5.5k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
App synch fails with ComparisonError rpc error: code = DeadlineExceeded desc = context deadline exceeded #3864
Comments
Have you tried to increase |
@WaldoFR thanks for the hint. I've added the following - --status-processors
- "20"
- --operation-processors
- "10"
- --repo-server-timeout-seconds
- "180" I am surprised as I only have 3 apps so far and I only sync them manually for now, so I would expect syncing one app should work and no timeout occurs |
Do you still face the synchronization problems ? |
No seems better now, apps show |
So it only temporarily worked - I got at least a status |
Ok, we face the same problem (in v1.5.3), and increasing number of
Seems that some other people are facing issue ( i.e. #3547 (comment) ) Related to this issue (#2957) may I ask you if you are currently deploying helm3 charts on your cluster ? It seems that we face this problems only on clusters with helm3 charts but we can't ensure it's related. Regards, EDIT : Additionnal informations, sometimes repository that are not syncing are synchronized for a few minutes. |
I have a single cluster and yes I have some stuff like cert-manager deployed with helm 3 |
If I switch back to v.1.5.0 I don't have the issue and sync work fine - updated the description with the new info. |
It's does not seem it's the version, I have now the same issue with |
So in my case is that I get a
|
I completely removed argocd and installed it in a new namespace. So far it seems good and I don't have the issue mentioned. Still not clear why I had it in the first place. |
Have you proceed to auto-sync same deployments of your previous ArgoCD setup ? |
@WaldoFR no at the moment I don't use auto-sync |
In my case I found that all commands were stopped at 15s exactly, browsing the code I found this part here that would explain we stop synchronization after 15 seconds :
Git client seems to not handle timeout specified in Regards, References : Line 112 in e92e0fa
|
I have the same issue with helm3, argocd 1.5.3 and cert-manager 0.16.1
|
and I have only dozen applications in control repo ... |
I have the same issue while adding a new https repository from bitbucket and gitlab: Having following error message: Unable to connect HTTPS repository: Get "https://bitbucket.org/xxx/yyy/info/refs?service=git-upload-pack": context deadline exceeded (Client.Timeout exceeded while awaiting headers) I enabled --insecure flag for argocd-server These are the settings applied to the argocd-application-controller:
I also tried to scale up argocd-repo-server (5 replicas) and it did not work. git clone inside argocd-repo-server is working fine. Installed following ArgoCD versions and had the same issue for all of them:
My cluster has 3 nodes and 1 master. This is happening for manual and auto sync policies while creating a new app. |
My argocd-repo-server logs:
I have got a bunch of no permission warning. |
These are no warnings. It's just a log of a command that got executed, which happen to be executed using a parameter |
Hi @cemsusal, is it happening when you create an app or when you add the repository? For the latter, an initial connection attempt is made by |
This is happening while creating and app or adding a repository from bitbucket or gitlab. (Did not try github) ARGOCD_GPG_ENABLED -> Is now set the false and issue still persists. We are not behind a proxy. argocd-server can access to the internet. |
This is a log from our repo-server:
This is confirming that repo server is failing to fetch. |
@jannfis this is the output of git commands:
|
I have only one application on top of a fresh install of v.1.7.1. The issue still persists. |
Installing Argo CD in another namespace did not resolve the issue for me. |
it may depends on this #3998 |
We're experiencing something similar in v1.7.6. Our application controller started spewing this log this morning:
The reason being our network is saturated and so packets are being dropped, connections timing out, etc. So while this kind of behavior is expected during a network issue, I'm concerned as to the impact its having on application syncs. What bothers me, is that the repo server hasn't thrown a single error and everything it needs is still in its cache. So why is the app controller timing out on syncs and putting a bunch of apps in the What is the point of the app controller hitting the Git repo with |
same problem with v1.8.3+0f9c684 |
Again, git timeout might be caused by slowness not related to ArgoCD. For instance the version control server you are using itself or tiers services related for authentication on it. To troubleshoot that, ensure that realizing a git clone from your node with sames credentials as you configured won't last over 15 seconds. |
What if the git timeout is definitely caused by slowness not (directly) related to ArgoCD ? To be more precise, what if the slowness is caused by rate limiting? Then argocd will start retrying all the calls that are rate-limited for more than 15 seconds, thus hammering the git server with more and more requests, thereby indirectly causing the rate limiting to increase. It would certainly be very useful if it were possible to change that timeout setting, in that situation. (PS: Sorry if this comes across emotionally, I just had a bad day trying to figure out why our production applications suddenly weren't updating any more, and why azuregit was suddenly refusing to talk to argocd.) |
thanks a lot man, i follow this and i fix my issue. before i googling 3 days to fixed it. |
We're experiencing the same timeout issues after deploying an ApplicationSet which creates 9 applications.
ArgoCD version deployed: v2.2.2+03b17e0 Controller settings: controller:
args:
# -- define the application controller `--status-processors`
statusProcessors: "60"
# -- define the application controller `--operation-processors`
operationProcessors: "40"
# -- define the application controller `--repo-server-timeout-seconds`
repoServerTimeoutSeconds: "360" GitHub API rate limits are 5000 API calls per hour, or ~83 calls per minute. I'm not knowledgeable enough about ArgoCD engine to say that we're hitting such but limits but I think it could be very plausible that such limits are being reached as I'm able to create the Applications one by one! UPDATE: I've also added # - define the application controller `--app-resync`
appResyncPeriod: "300" UPDATE 2: At the seconds refresh, the level=error msg="finished unary call with code Unknown" error="Manifest generation error (cached): `helm dependency build` failed timeout after 1m30s" grpc.code=Unknown grpc.method=GenerateManifest grpc.request.deadline="2022-01-25T14:18:48Z" grpc.service=repository.RepoServerService grpc.start_time="2022-01-25T14:12:48Z" grpc.time_ms=330.102 span.kind=server system=grpc |
@mnasruul what was the fix? |
follow this #3864 (comment) |
Maybe it could help to somebody as well. We have ArgoCD on AKS cluster and we faced similar issue. According https://argo-cd.readthedocs.io/en/release-1.8/operator-manual/high_availability/ we have tried to set workaround: |
Checklist:
argocd version
.Describe the bug
I've added different apps to my argo-cd but all of the fail to synch with the same error
To Reproduce
Here is the manifest from one of the demo projects which fails
Remarks
Not sure if this matters however...
rke
and contains 3 nodes.Expected behavior
Synchronization works fine and without errors
Version
Logs
argocd-server
argocd-application-controller*
I suspect the issue is cause if the
i/o timeout
For convenience, I've attached the logs from
argocd-server
andargocd-application-controller
The text was updated successfully, but these errors were encountered: