Fixbug: check driver timeout

[wnxn]

Please reference Issue #67.

[wnxn]

/assign @jsafrane

[jsafrane]

I am open to increasing probe timeout, however, I don't like the other changes.

[jsafrane]

I don't like 3 different contexts. If you think that 1 second is too short for GetDriverName+SupportsPluginControllerService+SupportsControllerPublish together, then we could increase the context timeout to 3 seconds instead of using three different ones.

[wnxn]

Hi, @jsafrane
Thanks for your reply.
After only increasing Probe timeout, I still meet timeout error in external attacher container. We may not call waitForDriverReady during the three calls, GetDriverName+SupportsPluginControllerService+SupportsControllerPublish.
If we don't use 3 different contexts, we can call waitForDriverReady before or after
the three calls.
When calling waitForDriverReady before the three calls, may waste some time because probe is not a short call.
On the contrary, we could call waitForDriverReady after the three calls and before launching CSI handler or trivial handler.

only increasing Probe timeout in external attacher:

# kubectl logs csi-neonsan-controller-0 csi-attacher
I0827 07:58:09.721682       1 main.go:75] Version: v0.2.0-35-g5a4cc3b-dirty
I0827 07:58:09.727437       1 connection.go:88] Connecting to /var/lib/csi/sockets/pluginproxy/csi.sock
I0827 07:58:09.727923       1 connection.go:115] Still trying, connection is CONNECTING
I0827 07:58:09.728326       1 connection.go:112] Connected
I0827 07:58:09.728351       1 connection.go:235] GRPC call: /csi.v0.Identity/GetPluginInfo
I0827 07:58:09.728363       1 connection.go:236] GRPC request: 
I0827 07:58:09.731362       1 connection.go:238] GRPC response: name:"csi-neonsan" vendor_version:"0.3.0" 
I0827 07:58:09.731484       1 connection.go:239] GRPC error: <nil>
I0827 07:58:09.731495       1 main.go:115] CSI driver name: "csi-neonsan"
I0827 07:58:09.731585       1 connection.go:235] GRPC call: /csi.v0.Identity/Probe
I0827 07:58:09.731629       1 connection.go:236] GRPC request: 
I0827 07:58:11.783301       1 connection.go:238] GRPC response: ready:<value:true > 
I0827 07:58:11.783424       1 connection.go:239] GRPC error: <nil>
I0827 07:58:11.783439       1 main.go:202] Probe succeeded
I0827 07:58:11.783453       1 connection.go:235] GRPC call: /csi.v0.Identity/GetPluginCapabilities
I0827 07:58:11.783464       1 connection.go:236] GRPC request: 
I0827 07:58:11.784170       1 connection.go:238] GRPC response: 
I0827 07:58:11.784301       1 connection.go:239] GRPC error: rpc error: code = DeadlineExceeded desc = context deadline exceeded
E0827 07:58:11.784501       1 main.go:125] rpc error: code = DeadlineExceeded desc = context deadline exceeded

[jsafrane]

Now I see what you meant. Can you move this check before ctx, cancel := context.WithTimeout(context.Background(), csiTimeout) ~10 lines above?

I think the order of waitForDriverReady() and GetDriverName() does not really matter, so let's do probe first and then start 1 context for GetDriverName+SupportsPluginControllerService+SupportsControllerPublish.

[wnxn]

Hi, @jsafrane
Currently, I move the probe check before GetDriverName+SupportsPluginControllerService+SupportsControllerPublish and increase probe timeout. Would you please review this PR again?

[jsafrane]

/lgtm
Thanks for the fix!

[jsafrane]

/approve
?

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: jsafrane, wnxn

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [jsafrane]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[jianglingxia]

@ jsafrane
Excuse me，in k8sv1.13 i create a pod with cinder-csi-plugin,the pod log is :rpc error: code = DeadlineExceeded desc = context deadline exceeded

Events:
Type Reason Age From Message

---

Normal Scheduled 1m default-scheduler Successfully assigned default/csi-pod-nodeworker to 111.111.111.116
Warning FailedAttachVolume 33s (x6 over 49s) attachdetach-controller AttachVolume.Attach failed for volume "pvc-13dc8ae1-61c8-11e9-b4f9-3a8d7c0bb20e" : rpc error: code = DeadlineExceeded desc = context deadline exceeded
Normal SuccessfulAttachVolume 17s attachdetach-controller AttachVolume.Attach succeeded for volume "pvc-13dc8ae1-61c8-11e9-b4f9-3a8d7c0bb20e"

the log print :rpc error: code = DeadlineExceeded desc = context deadline exceeded is that within 15s
not attached the node? this is a problem ?and how to fix it ? thanks very much!