Run ./pravega-segmentstore

[zzkww]

When I run ./pravega-segmentstore, I get an error. Why does this error occur?

[RaulGracia]

Hi @zzkww, we may need more details about how this cluster was configured and the circumstances of the error.
For example:

• Are you using Pravega standalone or not? If not, what storage binding have you used?
• Have you partially cleaned up the cluster? (e.g., uninstalled Bookkeeper but leave the existing data in long-term storage)

If you could share the reproduction steps, that would be helpful. Thanks!

[zzkww]

Hi @zzkww, we may need more details about how this cluster was configured and the circumstances of the error. For example:

• Are you using Pravega standalone or not? If not, what storage binding have you used?
• Have you partially cleaned up the cluster? (e.g., uninstalled Bookkeeper but leave the existing data in long-term storage)

If you could share the reproduction steps, that would be helpful. Thanks!

I deployed the environment in a k8s cluster, and initially I used s3 as the secondary storage. The s3 storage can upload those log metadata to s3. After that, I added another cloud service storage method similar to s3 to the source code of pravega to implement this function, and then redeployed zookeeper, bookkeeper, and pravega, then entered the controller pod and ran ./pravega-segmentstore This error occurred.

[RaulGracia]

Thanks @zzkww, so I understand you are adding your own storage binding to Pravega and testing if that works, right?

In that case, if you have been able to run Pravega with the S3 binding, the most likely case is that the new code for the new storage binding has some kind of bug. A couple of questions:

• Have you executed tests locally with the new storage binding on top of standalone instance of the new storage service you want to use? Normally, it is good to check if locally the binding works before running it against the cloud.
• Concretely, the error you are showing refers to Pravega not being able to find a file in the new storage service. Has Pravega being able to write any data at all in the new storage service? If not, could that be a credentials problem? If yes, what are the objects that have been successfully written?
• The call that is failing is doGetInfo(). Do you see other doGetInfo() calls to be successfull on other objects?

[zzkww]

Thanks @zzkww, so I understand you are adding your own storage binding to Pravega and testing if that works, right?

In that case, if you have been able to run Pravega with the S3 binding, the most likely case is that the new code for the new storage binding has some kind of bug. A couple of questions:

• Have you executed tests locally with the new storage binding on top of standalone instance of the new storage service you want to use? Normally, it is good to check if locally the binding works before running it against the cloud.
• Concretely, the error you are showing refers to Pravega not being able to find a file in the new storage service. Has Pravega being able to write any data at all in the new storage service? If not, could that be a credentials problem? If yes, what are the objects that have been successfully written?
• The call that is failing is doGetInfo(). Do you see other doGetInfo() calls to be successfull on other objects?

Thank you for your answer! I want to know how to write data to secondary storage (for example, writing data to S3)