drewg123

67 ms·

901.
by drewg123 5y ago
I just filled it and forgot it. Whatever I have it set to matches some medium-ish setting that I've selected because it "feels right" on concept-2 rowers when I travel.
902.
by drewg123 5y ago
How much water was in it? That's the only way to change resistance on the one that I have. How did you change the resistance for the one in the gym? FWIW, my resting heart rate is now in the 50s, and I get my heart rate up to betwee
903.
by drewg123 5y ago
Same comment but with WaterRower. The sensor measures the speed of the turbine in the water tank.
904.
by drewg123 5y ago
Exactly. For me its a rower, for others it might be a treadmill or a stationary bike or stair climber. The key is to make using it part of your routine. And, protip: If there are clothes hanging on it, you've failed.
905.
by drewg123 5y ago
I swear by indoor rowing. I've dropped 80lbs (from obese to my target weight) in the past 8 years, and most of that was from rowing I was running, but living in an area with seasons, there were just too many excuses (its too hot, its
906.
by drewg123 5y ago
Such sad news. I remember him from cdrtools. It was amazing because I recall that it ran everywhere in the late 90s. SunOS, Solaris, DEC UNIX, FreeBSD, Linux, etc. That was remarkable for a piece of software that interacted with hard
907.
by drewg123 5y ago
That's the obvious solution. But we A/B test QoE metrics to death. Touching something so fundamental about TCP would surely hurt QoE. The last time we tried to coerce TCP into helping with a problem like this, we were dealing wi
908.
by drewg123 5y ago
The problem is that with current generation NICs and NVME drives, you need a buffer somewhere, and there is no place for the buffer. You need a buffer because the NVME drive speaks in 4k chunks, and doing anything sub-4K is non optimal. So
909.
by drewg123 5y ago
If they could have worked out a deal to license MacOS X in the late 90s, it would have been a real game changer. Its too bad Jobs would not give Dell reasonable terms.
910.
by drewg123 5y ago
The feature you're talking about, DDIO, is worse than useless for our application. It wastes cache ways on I/O that has long, long, long since been evicted from the cache by the time we go to look at it. It might be helpful for
911.
by drewg123 5y ago
No, just multiple CX6-DX I'm currently stuck because the only boards we can find that have enough PCIe lanes exposed to do 800g (64 for NICs + close to 64 for NVME, with them more or less equally divided between sockets) have only 3 xg
912.
by drewg123 5y ago
Note that this was with the chip running in NUMA (4-NPS) mode. It did much better in 1-NPS mode when paired with a kernel without NUMA optimizations. Rome in 1-NPS mode is just fine, far better than naples.
913.
by drewg123 5y ago
This is my EuroBSDcon talk from 2 years ago. The more recent one (about getting to 400G) is up now at https://www.youtube.com/watch?v=_o-HcG8QxPc
914.
by drewg123 5y ago
All machines (can) run the same services, and we roll an entirely new self contained OS image for every update. That goes through multiple expanding levels of canaries prior to a fleet wide rollout. Bugs that escape the canaries are quit
915.
by drewg123 5y ago
We did this with FreeBSD several years ago at Netflix. We used to run heavily modified -stable kernels. Our tree was ahead of FreeBSD-current in a lot of ways (TCP, ktls, VM) but way behind in other ways. Every year or two we'd rebas
916.
by drewg123 5y ago
No. At least when I was there, the linux kernel was in git, and reviews were done in Gerrit. The monorepo (google3) and its review tools were used for most other projects.
917.
by drewg123 5y ago
Coming from Android, I used to use the Google assistant in hands free situations. Like when cooking or driving. I've found that Siri is TERRIBLE in these situations, because its more likely to punt to give you some links to click o
918.
by drewg123 5y ago
At least in the early 90s, most of the support programs on the commercial unixes were ancient BSD stuff from 10 years previous and just sucked. No command line editing, limited functionality, etc. This is why GNU became so incredibly popu
919.
by drewg123 5y ago
I admin'ed other machines at the time (DEC Alphas), but the guys who cared for the Sun systems ignored the package repo. I think it was because the packages were out of date. It also could have been because disk space was expensive,
920.
by drewg123 5y ago
Isn't it the same CPU, just differing numbers of cores with different clock rates and different TDPs? I'm assuming that at least part of this is just due to binning, where the best samples are 3Ghz high core count, and the lower
921.
by drewg123 5y ago
There is no context switch, like most system calls, sendfile runs in the thread context of the thread making the syscall. FreeBSD has "async sendfile", which means that it does not block waiting for the data to be read from disk.
922.
by drewg123 5y ago
We have servers in IX locations where there is a lot of traffic. Its not my area of expertise (being a kernel hacker, not a network architect), but our CDN load does not spill back to EC2. The biggest impact I have to QoE is when I crash a
923.
by drewg123 5y ago
We do all of our testing on production traffic, and we have servers in IX'es around the world.
924.
by drewg123 5y ago
No, nothing custom, we don't have access to the source. We like to run each server as an independent entity, so we don't run NVMEof. They're pretty much rock solid. If you'd like to discuss more, ping me via email. Use
925.
by drewg123 5y ago
I wish the video was online... We tried another vendor's NIC (don't want to name and shame). That NIC did kTLS offload before Mellanox. However, they could not retain TLS crypto state in the middle of a record. That meant we h
926.
by drewg123 5y ago
It depends. Normally about 10-ish percent. I've seen well over that in the past for super popular titles on their release date.
927.
by drewg123 5y ago
This is gone over in the talk. We allocate the page locally to where the data is used. The idea is that we'd prefer the NVME drive to eat any latency for the NUMA bus transfer, and not have the CPU (SW TLS) or NIC (inline HW TLS) sta
928.
by drewg123 5y ago
1) No. Well, potentially as a crypto accelerator, but QAT and Chelsio T6 are less power hungry and more available. GPUs are so expensive/unavailable now that leveraging them in creative ways makes less sense than just using a NIC lik
929.
by drewg123 5y ago
I recall that simple transmits from offset 0..N in a file worked. But range requests of the form N..N+2MB lead to corrupt data. Its been 2+ years, and I heard it was later fixed in Linux.
930.
by drewg123 5y ago
Async sendfile is also an advantage for FreeBSD. It is specific to FreeBSD. It allows an nginx worker to send from a file that's cold on disk without blocking, and without resorting to threadpools using a thread-per-file. The gist is

More ›