Closed fils closed 4 months ago
Hi @fils, the error seems to suggest that the database file is corrupted / not compatible with this version of Explorer. Does launching Explorer without mounting a local database works for you?
@mewim good catch, and that does run
[12:14:12.341] INFO (1): Access mode: READ_WRITE
[12:14:12.404] INFO (1): Version of Kùzu: 0.3.2
[12:14:12.404] INFO (1): Storage version of Kùzu: 26
[12:14:12.406] INFO (1): Deployed server started on port: 8000
What is the difference between "version of Kuzu" and "storage version of kuzu"
Also, it might be nice if that could be caught and reported as a error vs the "bad_alloc"
I'll try and align my kuzu versions between my code and explorer and report back.
@fils Did you create your dataset with a dev build of kuzu? For stable releases, each of them has a storage version which is also stored in the database catalog, so that we can check whether a database file is compatible with the version of kuzu and report an error. However, for dev builds, since it is built automatically by a CI job, the storage version is not maintained, so even if the file indicates it is the same storage version, the actual format maybe incompatible.
@mewim I bet this one was, was playing with some of the early RDF approaches and I know a fix got put in a dev build I wanted. I suspect this run environment still has it.
So, if it wasn't a dev build the check would have occurred? Good to know, and looks like it was all on me (go figure) ;)
Thanks for the help, I'll make the changes, re-run and report back.
@fils If you are using a dev build, you can also try to use the dev version of explorer by replacing kuzudb/explorer:latest
with kuzudb/explorer:dev
. For dev builds, it maybe a bit tricky to align the version of explorer and the version of dev build since we do not check automatically and only keep one latest dev build for the explorer app.
@mewim So it works now in docker via an Ubuntu VM running on my Fedora WS. However, I am still seeing an permission issue with the lock file on podman in Fedora itself. I can't work on this here now, but I will investigate and report back.
Thanks for your help! Have it working in docker at least. The people I want to share with will be using docker, not podman, so hopefully all is good now that the db versions are aligned.
Will update with what I find out re: podman
@fils Glad to hear it works for you in Docker. For podman, it seems that --userns=keep-id
did the trick for me when I tested it a few months ago. If it does not work now, I'll look into it again.
I am closing this for now. If you encounter another issue while using Explorer, feel free to reopen this or file a new issue.
Curious if there is a known reason I might be getting the following errors trying on on Fedora and Ubuntu.
Has this been seen before, a quick search of issues didn't seem to show any.
Thanks
podman on Fedora
same for Docker on Ubuntu
same issue on podman with the recommended