Zebra is down for hardware maintenance. Should be back online in a few hours max.
UPDATE: 5:40 PM EEST: Back online
Latest Entries »
Zebra server is down due to a damaged fiber optic cable. Should be back online in approx 24h. Will update.
Sorry for the inconvenience.
Update 05.23 9:30 AM EEST: Back online. Took a bit longer than I expected.
I’ve noticed that some people are trying to contact me by writing comments on my latest blog posts.
For my e-mail address, click the link below this post.
If you want to write me or ask something and prefer not to use e-mail, do write a comment under this post.
I want comments on other posts to be relevant to the theme of the post.
View Full Article »
I’ve decided to build myself a new PC for Christmas, to replace my aging i7 2600k.
lso got a new 4K monitor along with it.
Specs:
CPU: i7-9700k (cooled by Corsair H100i Pro)
RAM: 32 GB DDR4 3000 (4x 8 GB Corsair Vengeance LPX)
MB: Gigabyte Z390 Aorus Pro
GPU: GeForce 2080 Ti (Palit 2080 Ti GamingPro, TU102-300A chip)
Storage:
Samsung Evo 960 500 GB NVME SSD (from my old build)
Crucial MX500 1 TB SATA SSD
OCZ Vertex 3 120 GB SATA SSD (old beast, MLC based)
2x 2TB HDDs (Samsung HD204UI and Seagate ST2000DL003) in RAID 0
Case: Corsair 460X
PSU: Corsair TX850 V2 (old beast, but still going strong)
Monitor: LG 27UK600 4K @ 60 Hz
OS: Windows 10 Pro / Ubuntu 18.04 dual-boot
I plan to do some new Android x86 builds with this rig in near future. And of course, some 4K gaming.
— overclocking —
Tried doing some overclocking on the CPU and GPU, and I can already say that I likely did not win the highest prize in the silicon lottery. The 2080 Ti can handle +160 Core and +500 memory just fine but going beyond +160 core results in random GPU lockups (the more you push, the more unstable it gets) even though the temperature is way below 80 C. This is strange for what is supposed to be a higher-binned version of the Turing chip. The card is supposedly based on a reference schematic but with non-reference components. This suggests any BIOS from a reference card should work. And reference BIOSes do work (at least, they do not brick the card), but none I have tried so far are stable, even BIOS from OC version of the same Palit card.
As for the i7, I can’t find how to properly OC it. For my old 2600k, simply finding the right vCore voltage was enough, but squeezing out more power out of these modern chips requires much more research than just finding the right vCore. I did get 5 GHz to work, but with some random crashes every now and then. Right now, I’ve decided to call it quits and left the CPU at stock settings (I did leave the 95 W TDP limit disabled though).
Zebra is down due to power failure.
Update 18:00 EEST: back online.
Zebra server is down due to software issues, will try to fix as soon as possible.
EDIT: Sep 27th, 6:00 PM: Installed latest mainline Linux, all appears to work fine so far. Will update if anything goes wrong.
Raspis server will be moved to a new location today, expect downtime until everything is set up and the IP is updated.
Sorry for inconvenience.
UPDATE: Internet service at new location is not yet active, server move postponed until tomorrow.
UPDATE Sep. 20th: Server will be moved today. Hosts will be updated when the move is finished.
UPDATE Sep. 20th 6:45 PM EEST: Move is finished. Server back online and all hosts have been updated with new IP address.
Zebra server was down yesterday due to Ubuntu 18.04 update installing a buggy kernel which caused my primary network card to stop working. After spending several hours trying to figure why I can’t get an IP from my ISP, I’ve decided to switch back to my previous kernel. This fixed the problem.
It seems Ubuntu’s 4.15.0-33 kernel has issues with Realtek NICs. Strangely my other two NICs (both based on the same chip as the main one) were not affected.
More downtime is to be expected as I will be moving the server’s root filesystem to an SSD, probably tonight.
Zebra server was knocked offline by a power failure.
Power surge did take a victim as well – server’s power supply, a custom Dell OEM Lite-On model* got it’s standby circuit fried. It now only reads 2.5v (instead of 5v) when connected to mainboard. PSU does turn on without load but not with a mainboard connected. Sad. This was a good PSU, well made and very silent.
Well, at least the rest of the hardware seems intact.
I’ve temporarily installed an old 230W Hipro from an IBM Thinkcentre as a replacement. Won’t use it as a permanent replacement though, because the fan is way too loud.
Will have to go emergency PSU shopping.
EDIT: August 31st, 2018. One of servers hard drives started going south. It’s not the root FS drive, but since most of my servers 12 HDDs are of comparable age (except maybe for an 250 GB 3.5″ Samsung which is much older than the others) I have decided to get a cheap PCI-E x2 M.2 SSD and move my Linux root FS out of magnetic storage.
As for the new PSU, I’ve bought an 430W Seasonic S12II. It’s rating is overkill for my needs, but it’s dead silent, well built and has enough plugs to power my 8 SATA drives (other 4 HDDs are connected via USB. 2 internal and 2 external) without the need of any cable splitters or wiring mods.
* – my server is not a Dell, only the PSU was.
I will be moving Raspis server to a new location soon. I have not decided on a date yet, but it will probably be next week.
Will keep you updated.