Search results

did an esxi import woudn't boot - BSOD x7b or whatever it is - inaccessible boot device. tried the various disk controllers. the only one that would boot the system was the MegaRAID SAS 8708EM2. *however* I had loads of freakin' weird things going on. The os (server 2012 r2) booted fine, but I...

That is how bonding works. You won't see more than a 1gb/s for a given network stream (which, depending on hash policy could be source/dest mac, source/dest IP, source/dest IP and port, etc). Some protocols like multichannel smb can (theoretically) use multiple connections but it's something I...

ah, yes, of course. I have just ordered 3x960GB to use for special device. I think I will put them in a RAIDZ and steal some space for the boot/os. Maybe I will try small_files too. perhaps everytyhing under 512kb. Here is what I see: root@pbs:/mnt/datastore/RAIDZ/.chunks/aaaa# ls -lh total 94M...

My obvious reason for leaving out 'small files' is that, isn't every PBS backup chunk a fixed-size file? I thought they were all fixed at 4mb? none smaller, none larger?

It seems there's different comments on that space usage. Perhaps more people are doing 'small files' as well as metadata. I will try with what I have available anyway at first and keep an eye on it.

I made a small miscalculation.. I had in my head that my datastore is 36TB, but it's 44TB or therabouts (4x16TB RAIDZ). I still don't understand ZFS results in df but still at 0.3% for metadata it would be 132GB so those 200GB Intel enterprise SAS disks should be fine and I have 4 of them that...

Small files is optional isn't it? I plan to only put metadata on. 0.3% of my 36tb is about 108gb if I done my maths right.

OK. I have some enterprise 200gb SAS SSDs sitting around that I can use, and I can set up another PBS temporarily to shift the data off to and then back.

I have R7615 with 2x H965i. one controller has 1x 960GB Dell-branded Samsung PM9A3 drive, which I was forced to buy from Dell, plus 2x non-dell branded identical drives (so 3x in a RAID5, just to make it up to something useful with parity raid). The other controller has 4x Intel P5520 3.84TB...

Right. so 30 MB /sec is to be expected. Fine. but throughout this discussion it was said that the 12 MB /sec that I see when I restore to the older PowerEdge R630 is because of PBS HDD source. and I am saying how can that be, when I see 2.5* that performance when restoring to a newer PowerEdge...

Also, still no explanation why it is 30MiB/s to the newer target PVE node? admittedly, I have only run each restore once. Perhaps I will run again but in reverse order - second target node first, then the other one.

Thank you for the thorough reply! It is no problem for me to offload the data and recreate the volume. I just need to spend some time investigating required size of SSD special devices.

I believe atime has to stay active for garbage collection. There is some contradictory discussion on it, but if I understood correctly, it can maybe be changed to relatime but that's all, and it is already mounted as such: RAIDZ on /mnt/datastore/RAIDZ type zfs...

Ah, that is wonderful - perfect and so simple. Thanks very much. I will try it out this weekend.

I am testing out various recovery scenarios. One that I am testing, is to restore a disk image from a remote object storage (BackBlaze) which is mounted through kopia. After I get proxmox-backup-debug on to a pve host, I can successfully run the command: ~/proxmox-backup-debug recover index...

I think we're just waiting on a new driver package release from Redhat

That is correct in terms of where the data to restore lays. Thank you for your pointers and links. I have to go out now but I will do some study over the following week while I am away on holiday.

That was a restore to an older node (48 x Intel(R) Xeon(R) CPU E5-2670 v3 @ 2.30GHz (2 Sockets), target datastore 5x 1.92TB RAIDZ SAS 12G ZFS HBA330 no controller raid. I am now testing a restore to the original node which is a modern PowerEdge R7615 with the fastest single-core performance I...

I have a RAIDZ on PBS with 4x 16Tb SATA HDD. I am watching a KVM restore on a 10G LAN at 12MiB/s. Target pool is 5x 1.92Tb enterprise SAS 12g ZFS through HBA330 controller. how can it be so bad? I understand it's many small files, but 12MiB/s ? There must be a way to improve this? I read that...

yeah I know I linked to the GitHub issue, and it looks like there is some traction on fixing it finally. Fingers crossed :-)