text stringlengths 9 408 |
|---|
[917.48 --> 917.80] open. |
[917.94 --> 921.66] And it can be, if you've not done it before, quite an intimidating process. |
[921.66 --> 924.48] You're like, I've got $200 on my desk in front of me. |
[924.52 --> 925.38] I don't want to break it. |
[926.28 --> 927.14] I've heard a conspiracy. |
[927.68 --> 933.04] What I've been told is that they are drives that for some reason they suspect have lower |
[933.04 --> 933.96] quality to them. |
[934.04 --> 938.04] Like maybe the boards had QA issues or something like that. |
[938.40 --> 944.00] So they have the data that shows them that external drives don't run for as many hours |
[944.00 --> 945.50] as their internal drives do. |
[945.70 --> 947.14] And so they fail less. |
[947.42 --> 951.06] So they still think they're good enough for those use cases. |
[951.06 --> 954.24] So they package them up and they sell them off. |
[954.84 --> 959.82] And they figure it's good enough because those users generally will run them for 80% less |
[959.82 --> 961.38] time than the people that have them built in. |
[961.44 --> 962.38] So they'll never notice. |
[962.80 --> 964.44] I've heard that, but I don't know if that's true. |
[965.26 --> 968.22] I mean, we're never going to know if it's actually true or not, are we? |
[968.76 --> 969.66] But it would make sense. |
[969.76 --> 973.86] There'd be some MBA in an office somewhere worked out, how can we bin these drives that |
[973.86 --> 975.68] have failed a certain level of QA testing? |
[975.82 --> 976.06] Yeah. |
[976.42 --> 977.96] But you haven't really seen that bear out. |
[977.96 --> 981.04] I mean, you've used quite a bit of shucks drives and it's not like you have a lot of |
[981.04 --> 981.68] drive failures. |
[982.46 --> 982.96] Touch wood? |
[983.08 --> 989.22] No, I haven't had, I've had one actually fail, but I think that was a result of putting |
[989.22 --> 995.32] it into the Helios 64 because the data pins that snapped off and I had to solder and it |
[995.32 --> 995.88] was a bit messy. |
[996.00 --> 996.12] Yeah. |
[996.22 --> 999.60] So that was an actual Alex boo-boo, not a drive boo-boo. |
[999.60 --> 1005.84] That's anecdotal, but your anecdotal evidence would seem to suggest that it hasn't led to |
[1005.84 --> 1007.44] less reliable disks than I would say. |
[1007.74 --> 1008.60] So far, so good. |
[1009.10 --> 1009.36] Yeah. |
[1009.64 --> 1015.50] But then, you know, you've got to factor in that how much is a 14 terabyte drive on Newegg |
[1015.50 --> 1015.92] right now? |
[1015.96 --> 1017.78] I'm going to go and look it up right now, real time. |
[1017.78 --> 1026.06] So a comparable 14 terabyte drive on Newegg, a Western Digital Ultra Star, 14 terabyte drive, |
[1026.12 --> 1034.02] three and a half inch drive, $280 without the shell on it versus $200 at Best Buy for |
[1034.02 --> 1035.50] one with the shell on it. |
[1035.94 --> 1037.34] There are some warranty differences. |
[1037.54 --> 1038.78] This one has a five year warranty. |
[1038.94 --> 1042.18] I think the drives that come in the case only have three years. |
[1043.12 --> 1044.28] Factor that in as you will. |
[1044.28 --> 1052.14] But even so, I mean, once you do the calculations over enough disks, you actually end up saving |
[1052.14 --> 1053.16] quite a bit of money. |
[1053.32 --> 1059.44] So even if a couple more fail over time, you're still probably saving some cash. |
[1060.30 --> 1064.30] Now, do you find you're having to do many or any kind of tricks these days for the drives |
[1064.30 --> 1064.76] you're shucking? |
[1065.72 --> 1066.62] No, I don't. |
[1066.68 --> 1072.92] But that's simply because I built three years ago some custom SATA power connectors and |
[1072.92 --> 1075.58] I just snipped off the three volt rail in there. |
[1075.72 --> 1080.26] So, you know, whatever comes from my power supply may well have a three volt rail, but |
[1080.26 --> 1082.56] whatever I plug into the disks doesn't. |
[1082.74 --> 1084.02] So it doesn't matter to me. |
[1085.02 --> 1085.56] That is nice. |
[1085.74 --> 1086.44] That is nice. |
[1087.80 --> 1091.64] I'd like to mention our friends at A Cloud Guru have the Linux Web Services course. |
[1092.30 --> 1096.86] Learn the various software tools and utilities used to configure web services on a Linux host. |
[1096.86 --> 1101.68] You'll find out what the different use cases are for Apache, Squid, Nginx. |
[1102.16 --> 1107.00] You'll learn how to set up virtual hosts, proxies, when to use a reverse proxy, and how |
[1107.00 --> 1108.58] to manage SSL and load balancing. |
[1109.20 --> 1114.62] You'll also get some insights in Apache metrics and the differences between Apache and Nginx. |
[1115.20 --> 1119.90] We'll have a link in the show notes to this course or go to cloudguru.com and search for |
[1119.90 --> 1124.96] Linux Web Services, a cloudguru.com and the course Linux Web Services. |
[1126.96 --> 1132.66] So new drives, of course, mean that it's time to start figuring out where to plug stuff in |
[1132.66 --> 1138.84] and rebuild servers and, you know, just generally get my ducks in a row, right? |
[1139.00 --> 1139.28] Sure. |
[1140.20 --> 1146.38] So before I went to England, I threw in the new motherboard, the Asrock Rack motherboard |
[1146.38 --> 1148.70] that I wrote the blog post about and I mentioned on the show. |
[1148.70 --> 1149.98] It looked really good. |
[1150.10 --> 1152.52] I think we talked about it for a bit and I almost wanted to order one. |
[1153.02 --> 1153.14] Yeah. |
[1153.50 --> 1158.34] Well, unfortunately, I've discovered what I think is quite a big flaw in that motherboard. |
[1158.86 --> 1159.40] Uh-oh. |
[1159.92 --> 1162.50] Yeah, it's to do with how the PCIe lanes get shared out. |
[1162.66 --> 1167.72] So it doesn't sound like a big deal, except for the fact that I want to have about three |
[1167.72 --> 1176.04] or four different NVMe M.2 drives in there, as well as a PCIe U.2 NVMe drive as well, plus |
[1176.04 --> 1177.28] my HBA card. |
[1177.28 --> 1180.62] So by the time I've done that, I've used up quite a lot of PCIe lanes. |
[1181.60 --> 1185.84] And the real kicker was I tested something called PCIe bifurcation. |
[1186.50 --> 1192.66] And what that means is I can take the 16x slot and split it into three or four different slices. |
[1192.66 --> 1198.32] So I could take, I have a PCIe card that has four M.2 slots on it. |
[1199.32 --> 1206.14] And I can take my 16x basically graphics card slot and turn that into four M.2 slots. |
[1206.32 --> 1208.90] In this case, it was actually only three because of how the lanes worked. |
[1208.90 --> 1213.04] So when I did all the testing before I went to England, I plugged everything in and it |
[1213.04 --> 1213.58] all worked. |
[1213.68 --> 1214.56] And I was like, oh, brilliant. |
[1214.66 --> 1219.08] I can use this motherboard in my server, replacing one that had a lot more PCIe lanes that was |
[1219.08 --> 1220.58] a dual Xeon, et cetera, et cetera. |
[1221.14 --> 1226.50] Trouble was, I didn't actually put something in the fourth and final PCIe slot. |
[1227.22 --> 1228.34] Until now? |
[1228.68 --> 1229.14] What happened? |
[1229.72 --> 1230.36] Until now. |
[1230.36 --> 1232.18] Before I went to England, I did that. |
[1232.66 --> 1238.54] And I noticed that one of the NVMe drives wasn't showing up as part of my ZFS mirror that I |
[1238.54 --> 1240.68] have for my Docker app data. |
[1241.22 --> 1243.12] And I was like, oh, screw it. |
[1243.20 --> 1244.54] I'm getting on a plane tomorrow. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.