lpetrut's profile - activity

Just wondering, did you get to try out the latest MSI?

I ran some more tests just to be sure and got passed 4B WNBD requests and 2.5B RADOS requests: https://paste.openstack.o

Hi, Is it really a driver crash, as in a blue screen, or is it just that the driver disconnects the disks after detecti

The updated Reef MSI is ready, I've updated the top level answer with more details for better visibility.

Hi, Is it really a driver crash, as in a blue screen, or is it just that the driver disconnects the disks after detecti

Good news, I've found the issue and submitted a fix: https://github.com/ceph/ceph/pull/55413. We'll probably have an upd

This caught my attention: "op_send": 2147493692. Looks like a counter overflow on the rados side, I'm getting very close

A small update: hit the issue again after 3.5B requests, however we gathered more data this time. The good news is that

I've prepared a new test build and I'm attempting to reproduce the problem. I'm using memstore so hopefully it will show

Something that you mind find useful, I'm using WSL to interact with the running daemons: https://paste.openstack.org/raw

Got a few leads from the stack trace but it will need further investigation. It could be ceph messenger related or maybe

Good news, the issue reproduced after 5 days and 3.9B requests: https://paste.openstack.org/raw/b5xYFsPbZLXvftyyFlDf/. I

I need one more day to reach the 2 billion request mark :) https://paste.opendev.org/raw/bZK5NTBpxg441n8N3USi/. Once it

That’s a very useful observation, it might imply a counter overflow. I’ll look into it, thanks a lot!

Hi, Is it really a driver crash, as in a blue screen, or is it just that the driver disconnects the disks after detecti

Hi, How many events do you receive per minute, approximately? It may be worth checking the WNBD messages using either

Hi, The WNBD driver doesn't support Windows Server 2016. If you're only interested in cephfs, you can install Ceph with

For example: Windows host specs (number of cpu cores, total ram, available ram when hitting the problem), Windows versio

Could you share some info about the environment? It might give us an idea about possible limitations.

I couldn't reproduce the file copy error yet. How often does it show up? Are there any Dokan driver errors in the "Syste

About the unicode error, did you use the updated MSI? https://cloudbase.it/downloads/cephunicodereef.msi. Could you past

Thanks, we'll look into it as soon as possible.

Here's the updated MSI link: https://cloudbase.it/downloads/cephunicodereef.msi

Looks like a message decoding error. I'll rebase those patches against Reef and build another MSI to avoid messaging iss

Interesting. We switched to mingw-llvm, might be the reason why it didn't show up before. We have a patch that fixes ut

Interesting. We switched to mingw-llvm, might be the reason why it didn't show up before. We have a patch that fixes ut

Hi, The Windows driver hasn't been signed by Microsoft yet, which is why you need to disable Secure Boot as mentioned h

We'd need the event log messages as well in order to figure out what's wrong. Worst case scenario, the rbd-wnbd service

The above issues have been addressed, here's a beta Reef MSI that includes this fix: https://cloudbase.it/downloads/ceph

Also, the PR caches a single cluster connection, meaning that we can no longer connect to multiple clusters. One workaro

The PR hasn't merged upstream yet but we might apply it downstream. I did notice a few issues while running some tests:

Here's a preview MSI: https://cloudbase.it/downloads/cephdevsingle_process.msi. Please let me know if you hit any issues

Hi, I suggest checking the Openstack controller conf files (e.g. nova api) and use the same keystone endpoint. It might

Hi, Openstack Liberty was released in 2015, it was discontinued a long time ago. Also, we've announced that we'd no lo

The PR is up for review: https://github.com/ceph/ceph/pull/52540. Let me know if you'd like to try out a custom MSI befo

I've fixed the crashes, will have a PR soon. We'll most probably include this in the Reef msi.

I'll need to discuss it with the RBD team and see if it's a known issue.

The good news is that the OSD connections are shared when using a single process, so the number of connections is signif

Hi, Other Ceph users reported similar problems with rbd + qemu: https://ceph-users.ceph.narkive.com/SZb6qTQB/understand

That's expected. Openstack request headers must include the authentication token.

Hi, I suggest double checking the credentials from the [neutron] section. You can also compare it against your Openstac

Hi, I suggest double checking the credentials from the [neutron] section. You can also compare it against your Openstac

Thanks for the updates, keep us posted.

Hi, The MSIs from cloudbase.it are updated daily, including the latest WNBD version. You can verify it by doing a wnbd

Hi, The beta MSIs from cloudbase.it are updated daily, including the latest WNBD version. Regards, Lucian

Hi, What happens if you manually create that folder, i.e. by issuing a mkdir command? The following isn't necessarily

Hi, What happens if you manually create that folder, i.e. by issuing a mkdir command? The following isn't necessarily

Hi, For RBD mappings, there are two components: the WNBD driver and the userspace daemon (rbd-wnbd). rbd-wnbd logging

Have you specified the user? For example: rbd ls --user hostadmin.