Hello all!
I'm hoping that someone can help me out with this one. So about 2 weeks ago we had some storms roll through and knock out our power. No biggie as I keep my server on a UPS, so when the battery was getting low and it was apparent the power wasn't going to be back, I shut the server down. Crisis averted....kind of. The next day I went to power my server back on, and it did its usual thing, except I was then greeted by a grub boot failure. I booted into the installation media rescue mode and imported the pool to find that /dev/sda (part of my pool, but also the boot drive) was faulted. So I tabled getting it going, shut it down and ordered a new drive. I started following the directions listed here: https://oxcrag.net/2018/09/02/replacing-zfs-system-drives-in-proxmox/ But when I tried to create the partition table, I was getting errors again, so I figured I better dig deeper on this one. I opened the cover and found that apparently, when the server restarted and went to 100% fan speed (1U chassis, so it screams and everything is kind of crammed) it sucked one of my SATA cables back enough to hit the fan and broke one or two of the hair-like wires inside. So I shut it down and ordered a new cable. Now my configuration: Supermicro 1U X8DTU-F server and in it is an LSI 00301 SAS/SATA host controller card. No hardware raid. And then I have 4 2TB HDDs that make up the raid 10 (unfortunately, no extra drive bays or else I would have put the OS on its own mirror, then raid 10 the pool for the guest OSes).
At any rate, I followed the directions on that site again, created the partitions, and used dd to copy the data for the two partions. In the page, it showed copying partitions 1 and 9 (based on size, the 2 that are apparently not the data partition) and so I did the same, but using the numbers for my 2 smaller partitions. But this is where things diverged a little bit. When the page said to zpool replace rpool /dev/sda2 I got an error (sorry, I forget what the error is at the moment) I was able to get it to resilver though if I also included the drive's id in the command, but when I issued an lsblk, I saw that /dev/sda had only 2 partitions again while the mirror drive sdb had the 3 as shown earlier. (I assume /dev/sdb is the mirror - I assume there there is an easy way to verify that). Finally, when I tried to issue the command to install-grub, I was greeted with an error stating it "failed to get canonical path of 'none' " It seems from a search this may somewhat specific to the configuration/UEFI perhaps? But I can't say for sure. I'm fairly comfortable with Linux, but I never delved much into the hardware end of stuff with it, and I'm certainly a noob in terms of the grub bootloader (I've edited grub years ago on a dual boot desktop machine, but I've honestly forgotten more about it than I remember I think).
So, my request from those more knowledgeable than I - Would someone be able to walk me through this process start to finish? Of all the drives the fan had to chew the sata cable on, it had to be /sda with the boot loader of course :-/ But that seems to be how my luck goes lol I'd prefer to not have to reinstall from scratch again, and I view this as hopefully a learning experience. Just throwing in the towel and reinstalling doesn't teach me anything, even though at my current experience/knowledge base with proxmox, that would be the easy route for me hah
Thanks! I'm really enjoying proxmox so far!
Chris
I'm hoping that someone can help me out with this one. So about 2 weeks ago we had some storms roll through and knock out our power. No biggie as I keep my server on a UPS, so when the battery was getting low and it was apparent the power wasn't going to be back, I shut the server down. Crisis averted....kind of. The next day I went to power my server back on, and it did its usual thing, except I was then greeted by a grub boot failure. I booted into the installation media rescue mode and imported the pool to find that /dev/sda (part of my pool, but also the boot drive) was faulted. So I tabled getting it going, shut it down and ordered a new drive. I started following the directions listed here: https://oxcrag.net/2018/09/02/replacing-zfs-system-drives-in-proxmox/ But when I tried to create the partition table, I was getting errors again, so I figured I better dig deeper on this one. I opened the cover and found that apparently, when the server restarted and went to 100% fan speed (1U chassis, so it screams and everything is kind of crammed) it sucked one of my SATA cables back enough to hit the fan and broke one or two of the hair-like wires inside. So I shut it down and ordered a new cable. Now my configuration: Supermicro 1U X8DTU-F server and in it is an LSI 00301 SAS/SATA host controller card. No hardware raid. And then I have 4 2TB HDDs that make up the raid 10 (unfortunately, no extra drive bays or else I would have put the OS on its own mirror, then raid 10 the pool for the guest OSes).
At any rate, I followed the directions on that site again, created the partitions, and used dd to copy the data for the two partions. In the page, it showed copying partitions 1 and 9 (based on size, the 2 that are apparently not the data partition) and so I did the same, but using the numbers for my 2 smaller partitions. But this is where things diverged a little bit. When the page said to zpool replace rpool /dev/sda2 I got an error (sorry, I forget what the error is at the moment) I was able to get it to resilver though if I also included the drive's id in the command, but when I issued an lsblk, I saw that /dev/sda had only 2 partitions again while the mirror drive sdb had the 3 as shown earlier. (I assume /dev/sdb is the mirror - I assume there there is an easy way to verify that). Finally, when I tried to issue the command to install-grub, I was greeted with an error stating it "failed to get canonical path of 'none' " It seems from a search this may somewhat specific to the configuration/UEFI perhaps? But I can't say for sure. I'm fairly comfortable with Linux, but I never delved much into the hardware end of stuff with it, and I'm certainly a noob in terms of the grub bootloader (I've edited grub years ago on a dual boot desktop machine, but I've honestly forgotten more about it than I remember I think).
So, my request from those more knowledgeable than I - Would someone be able to walk me through this process start to finish? Of all the drives the fan had to chew the sata cable on, it had to be /sda with the boot loader of course :-/ But that seems to be how my luck goes lol I'd prefer to not have to reinstall from scratch again, and I view this as hopefully a learning experience. Just throwing in the towel and reinstalling doesn't teach me anything, even though at my current experience/knowledge base with proxmox, that would be the easy route for me hah
Thanks! I'm really enjoying proxmox so far!
Chris