dgentry - Network

Tomorrow should be zfs.rent initiation day! How fun!

Tomorrow should be zfs.rent initiation day! How fun!

Do I know anyone using Tello? Want to send me a referral code, before I sign up in the next couple days?

Do I know anyone using Tello? Want to send me a referral code, before I sign up in the next couple days?

Maybe a microsaas aimed at Europe... yeah, that would be a good thing right now.I'd need an idea though.Got nothin'.

Maybe a microsaas aimed at Europe... yeah, that would be a good thing right now.

I'd need an idea though.

Got nothin'.

I have officially entered information for my taxes this year. Please clap.

I have officially entered information for my taxes this year. Please clap.

USPS delivered the drive today.They rack new drives near the end of each month.

USPS delivered the drive today.

They rack new drives near the end of each month.

Espressif ESP32-S3-BOX-3 looks pretty neat. If I'm going to use these I guess I'd better order soon before tariffs make them...

Espressif ESP32-S3-BOX-3 looks pretty neat. If I'm going to use these I guess I'd better order soon before tariffs make them unaffordable.

https://heywillow.io/ appears to have enough functionality to get started.

So... maybe?

I think the drive is ready to ship to zfs.rent. I've done burn-in testing for a couple weeks, configured zrepl with it attached to a...

I think the drive is ready to ship to zfs.rent. I've done burn-in testing for a couple weeks, configured zrepl with it attached to a local VM, replicated all of the current filesystems, ran zpool scrub one last time, and removed it from the system.

It is back in the static bag, inside the bubble wrap cocoon, waiting to go to the post office tomorrow.

I don't even really remember what that project was. It suddenly became active again on March 16, and accrued charges since. Maybe I left...

I don't even really remember what that project was. It suddenly became active again on March 16, and accrued charges since.

Maybe I left something at the previous employer which could activate it again? Not sure why it started up, I don't recall doing anything which might.

Ok, great. AWS says I owe them money due to App Runner, but the console shows no App Runner services.So I get to figure out what this bill...

Ok, great. AWS says I owe them money due to App Runner, but the console shows no App Runner services.

So I get to figure out what this bill is really for, and turn it off.

I've concluded that this was indeed a network error which got magnified into a false array degradation event.That is honestly not cool...

I've concluded that this was indeed a network error which got magnified into a false array degradation event.

That is honestly not cool for ZFS to do, but I've decided to keep calm, use zpool clear, and carry on.

Hmm. https://openzfs.github.io/openzfs-docs/msg/ZFS-8000-9P/ says:"For example, the following cases will all produce errors that do not...

Hmm. https://openzfs.github.io/openzfs-docs/msg/ZFS-8000-9P/ says:

"For example, the following cases will all produce errors that do not indicate potential device failure: 1) A network attached device lost connectivity but has now recovered"

The kernel error messages all happened at 07:16:33 this morning.

07:16:33 is the precise moment that I sent a kill signal to the vzdump running on the other host, the Proxmox server where I'm using zrepl to replicate to this new system that saw disk errors.

I think maybe this isn't a hardware failure. It might just be the most terrifying possible way to report a network interruption.

Well, crappit. I ran badblocks tests for *weeks* on this device.Mar 31 07:16:33 zfsrent kernel: I/O error, dev sdb, sector 23368396833 op...

Well, crappit. I ran badblocks tests for *weeks* on this device.

Mar 31 07:16:33 zfsrent kernel: I/O error, dev sdb, sector 23368396833 op 0x1:(WRITE) flags 0x700 phys_seg 32 prio class 2
Mar 31 07:16:33 zfsrent kernel: I/O error, dev sdb, sector 23368399137 op 0x1:(WRITE) flags 0x700 phys_seg 32 prio class 2
Mar 31 07:16:33 zfsrent kernel: I/O error, dev sdb, sector 23368397089 op 0x1:(WRITE) flags 0x700 phys_seg 131 prio class 2
Mar 31 07:16:33 zfsrent kernel: I/O error, dev sdb, sector 23368401697 op 0x1:(WRITE) flags 0x700 phys_seg 3 prio class 2
Mar 31 07:16:33 zfsrent kernel: I/O error, dev sdb, sector 23368401441 op 0x1:(WRITE) flags 0x700 phys_seg 29 prio class 2
Mar 31 07:16:33 zfsrent kernel: I/O error, dev sdb, sector 23368399393 op 0x1:(WRITE) flags 0x700 phys_seg 200 prio class 2
Mar 31 07:16:33 zfsrent kernel: I/O error, dev sdb, sector 23368402721 op 0x1:(WRITE) flags 0x700 phys_seg 32 prio class 2
Mar 31 07:16:33 zfsrent kernel: I/O error, dev sdb, sector 23368402465 op 0x1:(WRITE) flags 0x700 phys_seg 32 prio class 2
Mar 31 07:16:33 zfsrent kernel: I/O error, dev sdb, sector 23368402209 op 0x1:(WRITE) flags 0x700 phys_seg 3 prio class 2
Mar 31 07:16:34 zfsrent kernel: I/O error, dev sdb, sector 23368401953 op 0x1:(WRITE) flags 0x700 phys_seg 32 prio class 2

root@zfsrent:~# zpool status
pool: pool1
state: DEGRADED
status: One or more devices has experienced an unrecoverable error. An
attempt was made to correct the error. Applications are unaffected.
action: Determine if the device needs to be replaced, and clear the errors
using 'zpool clear' or replace the device with 'zpool replace'.
see: https://openzfs.github.io/openzfs-docs/msg/ZFS-8000-9P
config:

NAME STATE READ WRITE CKSUM
pool1 DEGRADED 0 0 0
sdb DEGRADED 0 35 0 too many errors

D'oh!Ok, I'm currently looking for a house in Monterey in June sometime. I guess the first step in shaving the yak is to find an...

D'oh!

Ok, I'm currently looking for a house in Monterey in June sometime. I guess the first step in shaving the yak is to find an alternative to AirBNB with a good collection of possibilities there.

https://hachyderm.io/@SaareMartha@epicure.social/114142847340256220

Current offsite backup status:- gave up on Sanoid/Syncoid and went with zrepl instead.- zrepl successfully replicated three encrypted...

Current offsite backup status:
- gave up on Sanoid/Syncoid and went with zrepl instead.
- zrepl successfully replicated three encrypted partitions in raw mode.

However my first attempt on the first partition unintentionally replicated after first decrypting. I hadn't told the zrepl send side to use raw mode. This scares me. That first filesystem has nothing important in it, but it scares me that I'd made such a huge mistake and it wasn't at all obvious.

I *think* the subsequent three partitions replicated in raw mode are encrypted, and the remote machine doesn't have the key. However the error message it gives inspires less confidence:

`cannot mount 'pool1/backup/...': no mountpoint set`

So I'm now trying to doubly verify that the datasets are encrypted, and also verify that the backup worked and could be restored if needed.

Getting sanoid installed and regularly creating hourly/daily/etc snapshots was straightforward.syncoid, on the other hand... it doesn't...

Getting sanoid installed and regularly creating hourly/daily/etc snapshots was straightforward.

syncoid, on the other hand... it doesn't use the Sanoid snapshots at all, it makes its own? And it can maybe be started from cron, or not, you do you.

It seems like syncoid is a box of parts and I'm supposed to assemble the vehicle myself. After two days of googling, I'm still staring at the parts.

I'm looking at zrepl now. I just want an offsite backup.

Current offsite backup status:- added an encrypted dataset, instantiated new children for most of the existing filesystem and copied data...

Current offsite backup status:
- added an encrypted dataset, instantiated new children for most of the existing filesystem and copied data from the old unencrypted datasets to the new encrypted datasets.
- installed Sanoid, have it creating snapshots on a regular schedule and gradually purging older snapshots.
- set up a VM to approximate what the zfs.rent offsite backup will look like. I intend to send a drive already populated with the replicated files (though if they erase it during install I can create it again)

I'm working on Syncoid now, and I am confused.
- It doesn't use Sanoid's snapshots, it makes its own snapshot to replicate?
- It is supplied with Sanoid but doesn't run from Sanoid? I... run it from cron, or something?

I'm gonna do some more searching for a "How to set up Synoid for offsite backup" blog post. Surely the perfect writeup exists, I just have to find it.

I've had to start blocking people out of my life, or I'll never recover.Might still never recover.

I've had to start blocking people out of my life, or I'll never recover.

Might still never recover.

By the way: a zfs.rent like service with a datacenter in the EU would be extremely welcome right now.I've looked at rsync.net,...

By the way: a zfs.rent like service with a datacenter in the EU would be extremely welcome right now.

I've looked at rsync.net, hostbrr.com, layer7.net, pulsedmedia.com, NetCUP, tuxis.nl, and AlphaVPS. All provide the storage, and charge per TB.

I'm very fond of the zfs.rent model where I can trade capital cost for operating cost, but I haven't found any similar provider. I hope that doesn't mean the zfs.rent model is just not very viable, but maybe it does.

Soliciting strong ZFS opinions!The ZFS filesystems on my proxmox server, created years ago, are not encrypted. It wasn't a threat model...

Soliciting strong ZFS opinions!

The ZFS filesystems on my proxmox server, created years ago, are not encrypted. It wasn't a threat model I worried about back then.

I'm planning to ship a drive to zfs.rent next month, and now I do want the filesystems to be encrypted to be able to use zfs send in raw mode.

As I create the new filesystems, should I add them all under an encrypted dataset and inherit from the parent? Or make each one independently encrypted?

I can find blog post arguments for both. I'm wondering about downsides of hierarchal encrypted datasets.

We currently have a number of Google Home devices, one in the living room and a mini in each bedroom. We use them for:- sending broadcast...

We currently have a number of Google Home devices, one in the living room and a mini in each bedroom. We use them for:
- sending broadcast announcements like "Dinner!"
- setting alarms by voice command
- sometimes, playing music

Given what Amazon is currently doing with Alexa, I want to discontinue using the Google Home devices out of an abundance of caution.

I felt ok about Google's respect for privacy when we got them, but that was a different world and roughly a century ago or so.

I'm not sure there is a voice-activated open source or self hosted solution though. openVoiceOS looks more like a hobby than something one can just set up.