Xilinx Unified Installer 2020.1 -- Many days trying to download Vivado WebPack without success

[jkirwan]

I've been continually attempting to download the WebPack version of Vivado (for a Digilent Basys 3 board I just received) since Friday and over the weekend and Monday and now today without success. I do have a lot of experience getting it started, though. I'm able to get the unified installer to unpack itself just fine. That has never failed me. And once it starts up, I've attempted both what I imagine is an incremental download (which can be restarted from where it left off) as well as the full (not recommended) download. Neither method has worked for me and I've spend many, many hours with both methods. The error I always get (and rarely the same place twice) is "One or more files failed to download. Do you want to retry downloading them?" When I say "Yes," it appears to restart the download at the beginning but will, it seems, find that some of the content is already present and will move forward after a few seconds or minutes to a partial position before flipping me the indicated message again. I've seen it reach 90% or so of the download before quitting like this. But even after that much, it will still generate the error at 30% or 50% or elsewhere, too. (I'm computing that percentage -- the program doesn't actually report a number.) Sometimes, it might almost catch up to 90% again... but then fails. I have also looked over the other answers in "Community Forums/Forums/Vivado RTL Development/Installation and Licensing" which discussed problems similar to mine. I've turned off all of the Windows Defender Firewall stuff. None of that has helped or in any way changes the behavior. The machine I'm running on is a "work machine" at home. This just means it's my personal computer that I use for employment purposes (unrelated to any of this, as FPGAs are just a hobby to me) and that I keep absolutely pristine as to which sites I'll go to and to the software I'll download. It's almost completely vanilla Windows 10 Professional and it is maintained up to date by Microsoft in terms of their updates. And I have downloaded many, many gigabyte sized files before -- mostly from Microsoft's MSDN site for files that large -- without any problems. Not ever. Certainly nothing like this. I'm frankly stunned by the poor behavior because it's uniquely bad in my experience. Since everyone else I've downloaded from (Microchip, Microsoft, etc.) has zero problems and I've never, ever needed any kind of support getting an install completed, I'm just shocked seeing what I'm seeing here. I've fired up the Microsoft Resource Manager, Task Manager, and a variety of other tools by Microsoft to observe some of what's going on in the machine during these problems. One issue is that all 32 GB of my DRAM is fully occupied when running the installer. And, on a few occasions -- but not all by any stretch, I've observed extremely high page fault rates. Rates I've never seen before and never hope to see again. Much of the RAM is sucked up by Java, it appears, performing work for the installer. But I'm not certain, as that only comes from observation during installation. So don't hold me to it. It just looks that way to me. So here I am. No ability to download the software. Xilinx sales tells me to write here because they cannot help me. (I offered to provide a credit card so they could buy a memory stick, stuff the software on it, and send it to me.) I don't think (or at least, I'm not aware) there is any other avenue to getting the software. If there is a DVD of some kind, I'd certainly try that. But right now it doesn't look as if such exists. Is there any other way to get the software? Can I get it emailed to me? Can someone help me find what is causing the problems and causing this downloader to fail (unique among all my experiences in that it simply doesn't work at all?) After almost a 100 hours of babysitting this software, I'm about ready to find any other company's FPGA to practice my hobby on. I've never seen anything this bad in decades of experience. Seriously. It's that bad. And I've had bad experiences before. Just nothing quite like this. Thanks, Jon

[jkirwan]

Apparently, there's no one from Xilinx willing to write a response here. I'll re-write sales and complain.

[anatoli · Accepted solution]

Hello @jkirwan ,

Please note that there could be some sort of the internet drop out issue at your end, as many customers can download the software without any issues.

Since the webinstaller isn't an option, can you please tried downloading 2020.1 as a single file download?

Here is the link for this:

https://www.xilinx.com/member/forms/download/xef.html?filename=Xilinx_Unified_2020.1_0602_1208.tar.gz

Please try downloading this directly, and not via Unified WebInstaller. Please see if you can download 2020.1 that way.

Hope this helps.

[jkirwan]

I posted a long reply which appears to be missing. It appeared to be accepted when I posted it. Is there a process whereby my reply is being quarantined or something?

Thanks,

Jon

[jkirwan]

---

@anatoli wrote:

Hello @jkirwan ,

Please note that there could be some sort of the internet drop out issue at your end, as many customers can download the software without any issues.

I'm reposting my earlier reply, today. (Luckily, I have a habit of preparing all my posts with a separate tool that saves everything I write.) What follows is a copy. If it turns out to be a duplicate (the earlier one suddenly appears, later), when I'd appreciate reducing possible confusion by removing one of them.

---

Hello @anatoli. I'm fully aware that this is also a fact. An explanation must satisfy both that fact as well as the difficulties I'm having at my end and the fact that others (some of them long-time users of Xilinx parts) have also confirmed to me that they also experience these problems and haven't been able to resolve them.

Finally, an explanation must also take into account that every other software I've used, regardless of how it downloads to this particular machine over the last four years of my experience with it, has zero problems. I also perform a full scan disk on it each and every night. So that itself won't account for the unique problem here.

It's quite possible that the problems are in the internet path between your site and mine. There is a lot of software running on a lot of different machines, implementing who knows what variations on policies along the way, and therefore many opportunities for difficulties that are none of my fault and none of Xilinx's fault. If so, this may explain the facts as I think they are. Routing issues could certainly be a problem.

Here is my "tracert" while also running the WGET to fetch the Xilinx file:

Tracing route to a23-36-52-76.deploy.static.akamaitechnologies.com [23.36.52.76]
over a maximum of 30 hops:

1 7 ms <1 ms <1 ms 192.168.1.1
2 134 ms 23 ms 68 ms fdr01.grhm.or.nwestnet.net [50.46.181.58]
3 35 ms 43 ms 35 ms 50.46.176.78
4 47 ms 44 ms 15 ms ae2---0.cor01.bvtn.or.nwestnet.net [50.46.176.0]
5 12 ms 17 ms 30 ms ae0---0.cor02.bvtn.or.nwestnet.net [50.46.176.5]
6 43 ms 26 ms 32 ms cr2-bvtnorxb-a-be100.bb.as20055.net [204.194.220.5]
7 169 ms 53 ms 69 ms cr2-smrworxa-b-be14.bb.as20055.net [107.191.236.116]
8 10 ms 11 ms 38 ms cr2-sttlwawb-b-be12.bb.as20055.net [107.191.236.64]
9 69 ms 73 ms 62 ms cr2-sttlwawb-a-be10.bb.as20055.net [107.191.236.19]
10 27 ms 26 ms 27 ms 107.191.239.5
11 30 ms 53 ms 60 ms a23-36-52-76.deploy.static.akamaitechnologies.com [23.36.52.76]

Trace complete.

I note that WGET is using port 443 on [23.36.52.76] to fetch the file. If the port is wrong or malfunctioning or is handled oddly along the above route for reasons I cannot guess, then any of this could be a problem.

Here is an explicit error I'm getting: "decryption failed or bad record mac." This is happening only periodically. But it is happening continually, too. (The error occurs during SSL communications and apparently it's not able to be decided whether it was the decryption that failed or if the device MAC address didn't match up. Either way, it's an error and requires restarting.

Another error I got in the process, but it wasn't an explicit error, was the cessation of communication, entirely. The download simply stopped and didn't ever continue afterwards, once. In this case, at about 16 GB of download, WGET just sits there and lets me see that there's nothing happening and it happily reports it to me with a "--" as the download rate. So I can see that it stopped receiving packets, entirely. It will sit there for as long as I'm willing. In this case, I decided to wait about 15 minutes and then I aborted and re-started the download using "-c" as the option. This worked, perfectly, and the download rate was immediately back up to its normal rate.

So I've identified two different issues, here. I haven't identified the root cause of them. But I have been able to find out exactly what is transpiring between your server (or a server system you've hired to provide that service) and my location and machine. And these are not crippling problems for downloading to this machine. They are only minor errors, easily remedied. In one case, it's just a re-request. Trivial. In the other case, it's just an abort and re-start. Either way, the download continues again at full pace and without lasting impact.

Or it should be that way.

In the case of Xilinx download software, for reasons I won't belabor or argue about, it decides to restart the ENTIRE download process. It always starts at the beginning. However, it does seem to skip around -- by this I mean that while it always resets its download progress and starts over it also will seem to "touch base" at certain points, notice that there already some files downloaded, and will advance the progress bar at a speed that is far faster than my actual download rate. However, it also seems to "touch base" frequently along the way, so there are always continuing possibilities for this error. And when they occur, it's all the way back to the beginning again. For large downloads, and in the presence of these kinds of errors, it is just never able to complete the task. I think these are problems that should not cause Xilinx downloading software any lasting difficulties. It should just plow through them and move on. But it does NOT do that. Instead, it just fails and fails and fails -- for a week's worth of them in my case and without ever reaching any level of success in the end.

Now, in the case of the WGET software, it's a totally different result. One of brilliant success. I still get similar errors from time to time. However, in this case WGET has no problem re-synchronizing and re-starting the download at the exact correct position and in moving accurately forward without more than a moment's disruption of the download. It doesn't get flummoxed by such errors. It just moves forward, relentlessly, and gets the job done in short order, on time, and with all that you'd expect from a good downloading program in the presence of some kind of occasional error. In the case of the need for an abort, WGET also allows me to easily abort a download and re-start it. Which I did in that isolated case. Zero problems. It just works right.

This is the difference between well-written software and poorly-written software.

Well-written software will handle recovery, where possible, as well as possible.

Poorly-written software can, because it doesn't cope nearly as well in the presence of error events, become literally unusable.

---

@anatoli wrote:

Since the webinstaller isn't an option, can you please tried downloading 2020.1 as a single file download?

Here is the link for this:

https://www.xilinx.com/member/forms/download/xef.html?filename=Xilinx_Unified_2020.1_0602_1208.tar.gz

Please try downloading this directly, and not via Unified WebInstaller. Please see if you can download 2020.1 that way.

Hope this helps.

---

@anatoliThanks. While that's not an actual link -- it bounces me first to a page collecting information about me -- it does provide a direct link I can use afterwards. Which is great.

My sincere thanks,

Jon