43 replies to this topic

[WanWizard]

When you do an update, it first does an "opkg update".

 

This command downloads the package index file, Packages.gz, for all feeds, from the server. There are 7 in total, a few Mb's, which should be downloaded in seconds.

 

Once downloaded, each file is unpacked ( they are in gzip format ), and compared against what is installed locally, to determine if updates are available. This processing is all done on the box itself, which is why I asked to run the update from the commandline, so we can see the response.

 

The logs show however that instead of downloading all 7 in quick succession, there are ( 4 or more ) minutes between each request, meaning that either downloading the file takes very long, or processing the downloaded file on the box takes long.

 

As the logs show sub-second request times for each request, the problem is not server side, but either in transit, or on the box.

[dog-man]

When you do an update, it first does an "opkg update".

 

This command downloads the package index file, Packages.gz, for all feeds, from the server. There are 7 in total, a few Mb's, which should be downloaded in seconds.

 

Once downloaded, each file is unpacked ( they are in gzip format ), and compared against what is installed locally, to determine if updates are available. This processing is all done on the box itself, which is why I asked to run the update from the commandline, so we can see the response.

 

The logs show however that instead of downloading all 7 in quick succession, there are ( 4 or more ) minutes between each request, meaning that either downloading the file takes very long, or processing the downloaded file on the box takes long.

 

As the logs show sub-second request times for each request, the problem is not server side, but either in transit, or on the box.

 

Hmmm, perhaps the receiver is showing symptoms of old age, like me.

[WanWizard]

Age is between the ears... Most of us are approaching retirement, so we know all about that...

 

An Ultimo 4K is a powerful box, it should not have any issue with this. Have you checked the network connection of the box, it could be running in half duplex or incorrect speed due to a faulty cable for example?

[dog-man]

I have not checked the cable, but will do. I will first try doing an update via Wi-Fi, as it is currently connected via Ethernet cable to a power line adaptor or whatever they are called.

I do regularly power the adaptor down and back on.

It's not a cheap one, it's a Devolo one, and they are not cheap.

[WanWizard]

I have one of those connecting my home office to the switch, due to lack of ( time to install ) a cable.

 

Throughput absolutely sucks at times, even though it is a 1000mbps version, there are no phase issues in the UK ( virtually all houses are single phase ), and no idea what could generate interference.

[dog-man]

Is it worth me trying an update via command line before anything else?

[dog-man]

Link speed:     1 GBit/s
MAC address:     00:1d:ec:10:5d:9a
IPv6 address(es): 2001:4d48:ad5e:7e00:21d:ecff:fe10:5d9a/64 (SLAAC)
IPv4 address:     192.168.1.110/24 via 192.168.1.1 (static)

[WanWizard]

I would try the wireless connection first. I've got all boxes on wireless, as it is more reliable than powerline.

 

If you do

ifconfig eth0

you can see the TX and RX drops and retries, indicating a poor connection.
 

As to speed, the problem is that the connection between box and Devolo may be Gbps, the speed between both Devolo's is far from that.

 

Given the fact your internet connection is also Gbps, data arrives quickly, but then has to be squeezed through the slow powerline link, which may cause packet drops or connection disconnects, due to buffer overloads.

 

It like a 3-lane motorway, with a section of 2 closed lanes in the middle...

[dog-man]

I would try the wireless connection first. I've got all boxes on wireless, as it is more reliable than powerline.

OK, I will when I can get access. Family are hogging the room where the Ultimo is.

[dog-man]

The problem seems to be solved.

I switched to Wi-Fi and tried an update.

It worked very fast as it always did until recently and the brief blip a while ago.

No updates found, but it whizzed through the procedure.

[WanWizard]

Cool, thanks for the feedback.

[dog-man]

Well, that was short-lived.

The issue has returned this morning.

 

I checked, and wireless settings are fine within the receiver, nothing has changed.

I am thinking perhaps a router issue. Maybe my router is on its way out.

I don't think it is more than two years old, and it is a powerful router. It's a - GT-AX11000.

 

I can't say though, that I have noticed issues with any other gadget connected to my network.

Edited by dog-man, 15 August 2024 - 11:04.

[WanWizard]

Weird. You would say that if the router is to blame, you would also have issues on other devices...

 

I just did an update of a 9.0-release image, and that went well, so it is not an issue between the provider and our download servers.

 

Just thinking, you don't have IPv6 active on your network and box by any chance? We've had connectivity issues on IPv6 between the UK and OVH in the past. If I recall I saw SLAAC active.

 

You can check from the box using

ping downloads.openpli.org -c 1

and check if the response has an IPv4 or IPv6 address.

[dog-man]

Weird. You would say that if the router is to blame, you would also have issues on other devices...

 

I just did an update of a 9.0-release image, and that went well, so it is not an issue between the provider and our download servers.

 

Just thinking, you don't have IPv6 active on your network and box by any chance? We've had connectivity issues on IPv6 between the UK and OVH in the past. If I recall I saw SLAAC active.

 

You can check from the box using

ping downloads.openpli.org -c 1

and check if the response has an IPv4 or IPv6 address.

I do have IPv6 enabled on my router. No idea if it is enabled within the receiver.

If it is an IPv6 issue, why is it a recent issue and not a permanent one?

[dog-man]

I disabled IPv6 on my router, rebooted everything and all seems fine again.

I hope it lasts.

Strange, though, that IPv6 has been enabled for quite a while and the issue was not always apparent.

[WanWizard]

F'ups by OVH, in their network.

 

I've seen it before, massive packet loss on IPv6, I've complained about it, but they always come back saying "works fine for us, case closed". Absolutely horrible service.

 

It seems to be only a problem for users in the UK ( i.e. routing from the London IX to OVH ), I've received no complaints from other users.

 

You can force the box to use IPv4 by editing /etc/hosts, and add

51.178.16.136 downloads.openpli.org
51.195.116.98 downloads.openpli.org

 

[dog-man]

 

F'ups by OVH, in their network.

 

I've seen it before, massive packet loss on IPv6, I've complained about it, but they always come back saying "works fine for us, case closed". Absolutely horrible service.

 

It seems to be only a problem for users in the UK ( i.e. routing from the London IX to OVH ), I've received no complaints from other users.

 

You can force the box to use IPv4 by editing /etc/hosts, and add

51.178.16.136 downloads.openpli.org
51.195.116.98 downloads.openpli.org

Thank you, I will do that.

[dog-man]

 

F'ups by OVH, in their network.

 

I've seen it before, massive packet loss on IPv6, I've complained about it, but they always come back saying "works fine for us, case closed". Absolutely horrible service.

 

It seems to be only a problem for users in the UK ( i.e. routing from the London IX to OVH ), I've received no complaints from other users.

 

You can force the box to use IPv4 by editing /etc/hosts, and add

51.178.16.136 downloads.openpli.org
51.195.116.98 downloads.openpli.org

Should I delete the references to IPv6?

[Jork]

The IPv6 designers made a fundamental conceptual mistake: they designed the IPv6 address space as an alternative to the IPv4 address space, rather than an extension to the IPv4 address space.

[WanWizard]

The IPv6 designers made a fundamental conceptual mistake: they designed the IPv6 address space as an alternative to the IPv4 address space, rather than an extension to the IPv4 address space.

 

Not really an issue, in the 90's I used to manage a wide area network running IP and IPX together, that was a lot more fun.

 

The problem with UK --> OVH is packet loss, that has a technical issue, not a design flaw, at its core.

 

On and off this is been going on for months, it wouldn't be surprised if this is simply a capacity issue, not taking into account that most UK providers provide IPv6 now, and most client devices preferring 6 over 4.