rtp4me

While I recognize these are federal lands, I wonder if the states will pick up the slack to keep the parks open. In the end, Trump wants the states to do more with less federal government involvement, thus, I am not surprised at the force reduction. And, the layoffs seem to be happening for a while now (way before Trump).

For those of you who use "zrepl" for ZFS replication, you may want to checkout the fork from "dsh2dsh" - https://github.com/dsh2dsh/zrepl. This new fork provides a bunch of updates to zrepl including:

  * Local time-zones instead of only UTC
  * Faster replication jobs
  * Enhanced cron job definitions for pull/push jobs
  * New configuration for control and prometheus services

The original code works very well, but the new fork has added some enhancements the original author chose not to include.

Thanks for this! I have been using LXC/LXD for a long time and never knew about systemd-nspawn. Time to go learn something new!

We were running very large storage volumes in Azure (+2TB) and wanted to leverage ZFS compression to save money. After running some performance testing, we landed on a good balance of PGSQL and ZFS options that worked well for us.

It is - depending on the read-vs-write workload. For our workload, we landed on a record size (blocksize) of 128K which gives us 3x-5x compression. Contrary to the 8KB/16KB suggestions on the internet, our testing indicated 128K was the best option. And, using compression allows us to run much smaller storage volume sizes in Azure (thus, saving money).

We did an exhaustive test of our use-cases, and the best ZFS tuning options with Postgres we found (again, for our workload):

  * Enable ZFS on-disk compression

  * Disable ZFS in-memory compression (enabling this option costs us 30% perf penalty)

  * Enable primary caching

  * Limit read-ahead caching

Edit: Forgot to add, here are the required PGSQL options when using ZFS:

  * full_page_writes = off

  * wal_compression = off

Once the above options were set, we were getting close to EXT4 read/write speeds with the benefit of compression.

For our production PGSQL databases, we use a combination of PGTuner[0] to help estimate RAM requirements and PGHero[1] to get a live view of the running DB. Furthermore, we use ZFS with the built-in compression to save disk space. Together, these three utilities help keep our DBs running very well.

[0] https://pgtune.leopard.in.ua

[1] https://github.com/ankane/pghero

Sorry - just bought a new Toyota RAV-4; it does not have auto-lane change, the ability to auto merge onto the highway, the ability to auto-stop at the stop-light, etc.

It does have adaptive cruise control and the ability to stay inside the lines, but it absolutely will not automatically change lanes or exit off the freeway.

What? Every car can automatically change lanes, automatically merge into traffic, automatically exit to an off-ramp, automatically go around a round-about, etc? I think there may be a language barrier here because most cars/trucks I see on the road cannot automatically do any of these things.

Congrats to the team! I know it has been a long process, but kudos for sticking with it!

Same here. Turned in my 14" MBP M1 for a M2 15" MB-Air (1TB NVMe, 24G RAM). Lightweight, fast, excellent 15" screen. Hands down the best laptop I have ever had (and I have a stack of MBP boxes in my closet from the past 10yrs).

After seeing some of the unrealistic numbers, I set primarycache=metadata just like you pointed out. And, you are correct, I need to learn to be a better tester...

What methods/tools are you using to benchmark your ZFS systems?

We are currently testing a number of systems with 12x 30TB NVMe drives with Debian 12 and ZFS 2.2.0. Each of our systems have 2x 128G EPYC CPUs, 1.5TB of RAM, and a dual-port 100GbE NICs. These systems will be used to run KVM VMs plus general ZFS data storage. The goal is to add another 12x NVMe drives and create an additional storage pool.

I have spent an enormous amount of time over the past couple of weeks tuning ZFS to give us the best balance of reads-vs-writes, but the biggest problem is trying to find the right benchmark tool to properly reflect real-world usage. We are currently using FIO but sheer number of options (depth queue, numjobs, libaio vs io_uring) makes the tool unreliable.

For example, comparing libaio vs io_uring with the same options (numbjobs, etc) makes a HUGE different. In some cases, io_uring gives us double (or more) performance than libaio, however, io_uring can produce numbers that don't make any sense (eg: 105GB/sec reads for a system that maxes out at 72B/sec). That said, we were able to push > 70GB/secs large-block reads (1M) from 12x NVMe drives which seems to validate ZFS can perform well on these servers.

OpenZFS has come a long way from the 0.8 days, and the new O_DIRECT option coming out soon should give us even better performance for the flash arrays.

Maybe in your world, but certainly not mine.

To me, it is absolutely obvious. I have worked with lots of people in tech; mainly non-programmers (eg: lab managers, system engineers, cloud architects, etc). And, from my experience, many of those people prefer to talk face-to-face about detailed and difficult problems. Grabbing a marker and heading to a white-board to hammer out a tough problem is far better than a Zoom call with the drawing app. Can you do the same thing on a Zoom call? Sure, you can. But I think the results are much better in a face-to-face discussion.

And, to play devils advocate, how about the other way around? How about employers who really want employees to succeed but employees who don't want to put in the effort? Or, how about employees who believe what reddit tells them: they are being mistreated/abused because they can't have it their way?

Based on my own personal experience, good companies want to be accommodating because they understand people truly are the most important resource. They strive for a good balance of company-vs-employee desires. Who knows, maybe I am just lucky; maybe my past 30yrs working for big-tech is an anomaly?

I will be very happy when the vast majority of consumer vehicles are electric. The amount of extra heat generated by people sitting in their running cars with the AC running is incredible. I get it - this can’t be helped while at a stop sign or red light, but the sheer number of people who just park their car waiting to get their groceries (while on their phone), having lunch, sleeping, etc with the car running is absolutely ridiculous. So much extra heat needlessly generated.

If you are inclined to share your VC experience, you might want to reach out to the founder of Dioxus. As a recent YC 23 “winner” and recent (ex)employee of CloudFlare, I’m sure the founder would love to get your personal take on VC funding. I also understand they’re working on a product that CloudFlare could possibly use…

Indeed, it seems iVentoy uses NBD to present block storage devices (eg: ISOs) to the DHCP client. I just installed iVentoy 1.0.08 in my lab on a Debian 12 VM (via VMware 8 server). From there, I created a new VM and booted Linux Mint 21.1 via UEFI boot. Linux Mint install went well with no noticeable issues.

Overall, first impressions are looking good. Found a few bugs that need to get worked out including:

* UEFi boot requires iVentoy to have 2 vCPUs (see forums)

* Restarting the iVentoy script resets the IP Boot config (specifically the Subnet Mask and Gateway)

From their release page, iVentoy supports a bunch of Linux and Windows install images, and you can even inject custom auto-install scripts to the ISO for unattended installs. Very cool. Finally, the iVentoy discussion forum seems very active, and the developer seems engaged.

I will probably support the developer ($49) because he was able to leverage NBD to overcome some iPXE issues I struggled with for a long time. I know how much time/effort goes into making iPXE booting look seamless. Kudos to him/her/them

Personally, I don’t think you’re in the minority at all. I suspect many people prefer working in an office collaborating with their peers (myself included).

My $0.02. Probably not much different...

In my opinion, it's just the "human condition". Like adults, kids want to take the least path of resistance. Since learning is difficult for some/most of us, avoiding school is the easiest path since it requires real effort. This is why parental guidance is so important for young people. Sure, some cities experience violence, some schools have odd teaching mandates, etc, but all-in-all, it's the daily grind that kids want to avoid (just like adults!). Not to mention the pressure from social media.

While not quite hackable, I just bought (today) a Minisforum UM773 with 32G DDR5 RAM, 512G NVMe, and an AMD Ryzen 7 7735HS (8C/16T - 3.2GHz base clock). Paid $579 shipped to my door. Both the RAM and NVMe drive are easily upgradable. Gonna use it as an ESXi lab node.

Personally, I would find this very useful. I (used to) create internal tech training videos for our organization and would routinely stumble when doing voice overlay. Even though everything was scripted out, it took lots of editing time to get the audio and video aligned without the vocal stumbling (ahs, ehs, silence, voice redos, etc). Just my $0.02

We have this sort of issue with my mother-in-law (90s). The best thing I did was setup a Chromebook (ChromeOS Flex) with just the basics (mail, photos, etc). From there, we routinely login to the GMail account and remove crap from her emails. It is tough - especially since the world is so on-line (photos of grandchildren, etc). Given the huge disparity between her age and technology, it is extremely important we act as her proxy for her online communications. Privacy be dammed.

Thanks for this. Today, I had to open an account to send money to a friend. I was torn between Zelle and Venmo. Per my son's recommendation I went with Venmo. Not sure how many folks get screwed by Zelle, but your post made me feel a little better about going with Venmo. Although, I am sure Venmo has its fair share of shady characters...

Yep - 100% agree. I opened a new business back in Nov/Dec 2022, and no less than two weeks did I get multiple, official-looking payment requests for labor law posters. They looked very real with some verbiage about "failure to comply with the law", etc. After I opened the first one, I was very motivated to pay the $89 fee (thinking I missed something when starting the business). However, after reading the letter a few times, I realized it was a scam.

The others went in the trash as soon as they came in...

No kidding. It would be fascinating to see who joins the bidding process and what kind of leverage/collateral they use.

Slightly tangental, Fidelity also has the option to put your accounts in "lock down" mode to prevent funding *outflows* (even transfers between accounts). This can help secure your account in addition to 2FA. I learned about this recently after +10yrs at Fidelity. Sadly, I don't think it is enabled by default.

I wish my bank had the same sort of option to prevent someone from just randomly guessing my account number for an ACH transfer.

420 - up in smoke

Wow - vampire taps. That brings back some memories :-)