Search results

No, in fact it is not working with ntfs deduplicated volumes. Some pve versions (depending on in kernel ntfs driver) gives an error when trying to restore, and latest versions gives no error but restored file is empty! And don’t forget : dedup tasks on Windows volumes starts a few days after...

Yes, we worked on that but at the PVE host level, and concluded everything works better with ntfs3 (given by Paragon Software for Linux 5.15?). Everything works well EXCEPTED deduplicated files, despite some notes about it (https://www.paragon-software.com/ntfs-deduplication-support/#features)...

Perfect. Thanks.

Hi, Thank you very much for clarification. I need some more : suppose a chunk belongs to 50 VMs. And a verification task begins (datastore or namespace level). How many times will this chunk be verified ? Christophe.

Well, it was a perfectly working but misplaced traffic control, sorry... Christophe.

We didn't tried iperf when the problem occurred. Now it is 939Mbits (on a gigabit link : quite normal).

Well, problem seems solved by adding an IP alias in same subnet than the main one (/21), then bench via this alias was ok!! Tested again on main IP : ok! No more bad test, even after a reboot. I have NO idea where does the problem came from, still there after 2 reboots... Christophe.

No, nothing. 2 NICs. - One quad port I350 Gigabit Network Connection (1Gbps) - One dual port FastLinQ QL41000 Series 10/25/40/50GbE Controller. No, tests aree now done locally from PBS to PBS, via localhost, or IP. Via localhost : 390MB/s Via "real" IP : 1,20MB/s. nothing special about this...

Hi, I don't know if this is still true, but we have 2 x 3 nodes PVE 8 clusters + PBS with lots of timeout, both built on same hardware, and by the same guy. Both clusters suffers abyssal TLS perf a soon as traffic goes through network stack. Something seems wrong with TLS : [root@pbs1:~]#...

So nobody was really aware... Well, no problem : it works! Christophe.

Hi all, It seems PBS 3.0 is able to file-restore files from a deduplicated ntfs Windows 2022 VM. This is great! PBS 2.x was NOT able to do so. I haven't seen this feature in any doc yet, but more and more of our users are happy. Is someone from proxmox team aware of that? ;) It seems PBS 3.0...

Yes. It is enough to see VG. As soon as multipathing is working, you can hide iSCSI part of the storage. But in the end, this is NOT what we did : keeping them in proxmox GUI gives an easy way to know a path is down. To hide them, iirc, edit /etc/pve/storage.cfg is the way. Regards...

Yes, it is my understanding. In my case it's a move disk from a Ceph storage to an iSCSI one. Best solution for me : chose an async_io : threads, which seems to live together with any cache policy and any storage type. Regards, Christophe.

Why? Native involves no cache, which is not necessarily the best cache policy. Threads seems more flexible. Regards, Christophe.

Hi all, This kind of error is annoying... A default option generating errors is not an acceptable option, at least for us. Can we hope someone will solve the problem? If not, I will change on our clusters from io_uring to native. Don't know performance gain / loss but with our workloads...

Well, iSCSI is not PVE specific, and it is easy to connect a LUN in web gui. multipath is covered by documents, if if you let it « as is «, keep your iSCSI storage and your LVM storage based (« base » option in storage.cfg ), you are on the wrong way. on the other side, when you delete an...

Yes, It is exactely the point. I have never seen this clearly written in a global official PVE howto. Something is missing here in official docs about iSCSI, multipath, and integration in PVE. About your « base » option comment : none, but one by path could be useful to have a view in web ui...

Hi all, Really interesting thread. Shouldn't default values full_ratio (0.95) and nearfull_ratio (0.85), which are very aggressive numbers adapted for a test cluster (https://docs.ceph.com/en/quincy/rados/configuration/mon-config-ref/#storage-capacity), be set to a more realistic value at...

Well, not really, at least for me. On a 7.1-12 PVE cluster, 3 nodes, full flash ceph storage. 3 VMs. Windows 2019 VMs. Exactely the same, excepted saize of data disk. 2 of them have never had a problem during backup. Third one has same issue (backup mode = snapshot), nealy 2 days on 3... Same...

Hi, Same problem here. It seems rebooting every node after a kernel upgrade solves or at least mitigates the problem. Not clearly, from my point of view… Christophe.