Problem with long file stores ID in JVM options

[fredericbeck]

What steps does it take to reproduce the issue?

• When does this issue occur?

When adding a file store with a long name in the JVM options, file uploads fail whatever the upload method used, generating a javax.ejb.EJBException: String index out of range: -3 error in the server logs.

Everything works for short ID, e.g. sed-nge-storage or dv-global, but not for longer ones such as dataverse-global-storage

• Which page(s) does it occurs on?

Error occur on the file upload page, or when using DV Uploader or direct curl API access.

• What happens?

File upload fails, detailed error log can be found here.

• To whom does it occur (all users, curators, superusers)?

All users are impacted.

• Workaround

Using shorter file store IDs works fine

Which version of Dataverse are you using?

Using Dataverse v5.8

[qqmyers]

@fredericbeck - thanks! Having looked after chat yesterday, I'm fairly sure that the problem with the line @pdurbin referenced -

dataverse/src/main/java/edu/harvard/iq/dataverse/dataaccess/S3AccessIO.java

Line 181 in c753921

if(!storageIdentifier.substring((this.driverId + "://").length()).contains(":")) {

is that it should just be if(!storageIdentifier.contains(":")) { .

I think this is 'just' a bug - a cut/paste error. It will cause problems whenever the length of the store name is within a couple characters of being longer than the bucket name.

[pdurbin]

@fredericbeck are you interested in creating a pull request for this? We don't want to deprive you of long and meaningful IDs for stores. 😄

[pdurbin]

@qqmyers to be clear, we're saying this only affects the S3 driver, right? The file system driver should be ok?

[AR-2910]

Hey @pdurbin! I would like to address this issue and implement the solution provided by @qqmyers. Could you please assign it to me?

Regards

[pdurbin]

@AR-2910 it looks like @qqmyers assigned you. Thanks!

[AR-2910]

hey @pdurbin @qqmyers ,
I've reviewed the code related to this issue. I noticed that the specific line to change is different from what the issue originally asked for. The line in question is as follows:
if (!storageIdentifier.substring((this.driverId + DataAccess.SEPARATOR).length()).contains(":")) {

This differs from the proposed change in the issue description. The issue initially requested changing it to:
if (!storageIdentifier.contains(":")) {

The current code appears to have evolved, and the context of the check has changed. To address the issue and align with the current codebase, I plan to implement the proposed change line that does not involve substring. This change aims to simplify the code and make it more robust

Before proceeding, I wanted to confirm the correct line to modify. Could you please provide clarification on which change should be made?

Regards

[qqmyers]

Looks like that line moved to

dataverse/src/main/java/edu/harvard/iq/dataverse/dataaccess/S3AccessIO.java

Line 194 in 0af5caf

if(!storageIdentifier.substring((this.driverId + DataAccess.SEPARATOR).length()).contains(":")) {

[AR-2910]

Hey @qqmyers @pdurbin,

Thank you for assigning issue #8312 to me. I appreciate the opportunity to contribute to this project.

I have created a pull request #10031 for the code change.

Please review the pull request at your convenience. If the changes look good and meet the requirements, I would appreciate you merging it into the main branch.

Regards