network_connect_update-20200427, SNS 2.3.2 & NetWiz 2.1.3

[rerwin]

Uploaded new versions of simple_network_setup (2.2) and network_wizard (2.1). They contain several refinements and a SNS bug fix that now allows SSIDs longer than 20 characters. Thank you, to chili dog for his report and assistance with the fix.

[rerwin]

Uploaded improved versions of SNS (2.2.1) and Network Wizard (2.1.1) to correct and expand the changing of the "current network tool", so that the change is made only if a connect/disconnect or interface button is "clicked".

When SNS detects an active connection and determines that it was not started by SNS, it reports that and advises disconnecting and re-connecting, so that a re-boot will connect by using SNS.

[rerwin]

Uploaded SNS 2.3 which corrects waiting for modules to load and for connections to complete.

Thank you to charlie6 for reporting the issues and testing my fixes.

Now the b43 driver should work and slow connections should succeed.

[rerwin]

I have belatedly uploaded network_connect_update-20200410-beta to match the latest SNS and Network Wizard packages. It went into woofCE over a year ago and should have been added to the first message of this thread for older and non-woofCE distributions. Apologies.

The beta traces the logic of network initialization, to assist in troubleshooting reports of failure to automatically reconnect at boot-up time, in BionicPup64.

[LateAdopter]

Hello rerwin

I did the following:

Create new copy of Bionicpup64 in menu.lst
Do UK locale stuff
Create savefile
Install updated SNS, NetWiz and network_connect_update-20200410-beta.pet
Reboot, just in case..
Save bootsysint.log
Choose SNS in Internet Connection Wizard

Usual error: There does not seem to be any network connected to eth0

Run /usr/sbin/dhcpcd

Network connects OK

From Messages eth0 bit is this:

Code: Select all

Apr 11 09:28:39 puppypc9645 kern.info kernel: Generic FE-GE Realtek PHY r8169-2500:00: attached PHY driver [Generic FE-GE Realtek PHY] (mii_bus:phy_addr=r8169-2500:00, irq=IGNORE)
Apr 11 09:28:39 puppypc9645 kern.info kernel: r8169 0000:25:00.0 eth0: Link is Down
Apr 11 09:28:39 puppypc9645 kern.info kernel: Generic FE-GE Realtek PHY r8169-2500:00: attached PHY driver [Generic FE-GE Realtek PHY] (mii_bus:phy_addr=r8169-2500:00, irq=IGNORE)
Apr 11 09:28:39 puppypc9645 kern.info kernel: r8169 0000:25:00.0 eth0: Link is Down
Apr 11 09:28:39 puppypc9645 kern.info kernel: Generic FE-GE Realtek PHY r8169-2500:00: attached PHY driver [Generic FE-GE Realtek PHY] (mii_bus:phy_addr=r8169-2500:00, irq=IGNORE)
Apr 11 09:28:39 puppypc9645 kern.info kernel: r8169 0000:25:00.0 eth0: Link is Down
Apr 11 09:28:44 puppypc9645 kern.info kernel: gsettings-data-[4800]: segfault at 48 ip 00007f9af47ed05b sp 00007fff7c9235e0 error 4 in libgconf-2.so.4.1.5[7f9af47cf000+2e000]
Apr 11 09:28:44 puppypc9645 kern.info kernel: Code: 56 41 55 4d 89 c6 41 54 55 49 89 d4 53 48 89 f5 48 89 fb 41 89 cd 48 83 ec 20 64 48 8b 04 25 28 00 00 00 48 89 44 24 18 31 c0 <48> 83 7f 48 00 c7 04 24 01 00 00 00 c7 44 24 04 00 00 00 00 74 0b
Apr 11 09:30:12 puppypc9645 user.notice root: /usr/sbin/pkg Repo files updated by  1.9.22
Apr 11 09:32:40 puppypc9645 kern.info kernel: Generic FE-GE Realtek PHY r8169-2500:00: attached PHY driver [Generic FE-GE Realtek PHY] (mii_bus:phy_addr=r8169-2500:00, irq=IGNORE)
Apr 11 09:32:40 puppypc9645 kern.info kernel: r8169 0000:25:00.0 eth0: Link is Down

After running dhcpcd this is added to messages

Code: Select all

Apr 11 09:35:58 puppypc9645 daemon.info dhcpcd[6889]: version 6.6.2 starting
Apr 11 09:35:58 puppypc9645 kern.notice kernel: cfg80211: Loading compiled-in X.509 certificates for regulatory database
Apr 11 09:35:58 puppypc9645 kern.notice kernel: cfg80211: Loaded X.509 cert 'sforshee: 00b28ddf47aef9cea7'
Apr 11 09:35:58 puppypc9645 daemon.warn dhcpcd[6889]: all: IPv6 kernel autoconf disabled
Apr 11 09:35:58 puppypc9645 daemon.info dhcpcd[6889]: eth0: adding address fe80::1b46:f7cc:3777:1107
Apr 11 09:35:58 puppypc9645 daemon.err dhcpcd[6889]: if_addaddress6: Operation not supported
Apr 11 09:35:58 puppypc9645 daemon.err dhcpcd[6889]: script_runreason: /lib/dhcpcd/dhcpcd-run-hooks: WEXITSTATUS 1
Apr 11 09:35:58 puppypc9645 daemon.warn dhcpcd[6889]: no interfaces have a carrier
Apr 11 09:35:58 puppypc9645 daemon.info dhcpcd[6889]: forked to background, child pid 6936
Apr 11 09:35:58 puppypc9645 kern.info kernel: Generic FE-GE Realtek PHY r8169-2500:00: attached PHY driver [Generic FE-GE Realtek PHY] (mii_bus:phy_addr=r8169-2500:00, irq=IGNORE)
Apr 11 09:35:58 puppypc9645 daemon.info dhcpcd[6936]: eth0: waiting for carrier
Apr 11 09:35:58 puppypc9645 kern.info kernel: r8169 0000:25:00.0 eth0: Link is Down
Apr 11 09:36:05 puppypc9645 daemon.info dhcpcd[6936]: eth0: carrier acquired
Apr 11 09:36:05 puppypc9645 kern.info kernel: r8169 0000:25:00.0 eth0: Link is Up - 1Gbps/Full - flow control rx/tx
Apr 11 09:36:05 puppypc9645 daemon.warn dhcpcd[6936]: all: IPv6 kernel autoconf disabled
Apr 11 09:36:05 puppypc9645 daemon.err dhcpcd[6936]: script_runreason: /lib/dhcpcd/dhcpcd-run-hooks: WEXITSTATUS 1
Apr 11 09:36:05 puppypc9645 daemon.info dhcpcd[6936]: DUID 00:01:00:01:26:24:4c:05:00:d8:61:c8:cd:b1
Apr 11 09:36:05 puppypc9645 daemon.info dhcpcd[6936]: eth0: IAID 61:c8:cd:b1
Apr 11 09:36:06 puppypc9645 daemon.info dhcpcd[6936]: eth0: soliciting a DHCP lease
Apr 11 09:36:08 puppypc9645 daemon.info dhcpcd[6936]: eth0: offered 192.168.1.5 from 192.168.1.1
Apr 11 09:36:14 puppypc9645 daemon.info dhcpcd[6936]: eth0: leased 192.168.1.5 for 86400 seconds
Apr 11 09:36:14 puppypc9645 daemon.info dhcpcd[6936]: eth0: adding route to 192.168.1.0/24
Apr 11 09:36:14 puppypc9645 daemon.info dhcpcd[6936]: eth0: adding default route via 192.168.1.1
Apr 11 09:36:14 puppypc9645 daemon.err dhcpcd[6936]: script_runreason: /lib/dhcpcd/dhcpcd-run-hooks: WEXITSTATUS 1

I have attached the bootsysinit.log. (real gz!)

[rerwin]

LateAdopter,

Thanks for the log. I found the cause!

The network-wizard package prevents the detection of eth0, because it erroneously indicates that network-wizard is the current network manager even though it has not been set up.

The package contains a dummy file from woofCE that normally gets deleted in the distro build process.

After installing the network_wizard-2.1.1 package, delete the file:
/etc/network-wizard/network/interfaces/EMPTYMARKER

If you use the same setup, also delete any file in /etc/simple_network_setup so that the default eth0 can be detected.

[LateAdopter]

Hello rerwin

I deleted the file and the bootsysinit.log looks plausible, but SNS still fails. I notice that it gives up after about 6 seconds.

[rerwin]

Thanks. The log shows that the eth0 interface did not actually connect. It tries 5 times but was not successful. The log does not include the attempt to reconnect to SNS, but that is understandable since it would be a different reboot.

Did you have an ethernet cable plugged in and available for connection? If so, we need to play with the wait period. If not, that would explain the timeout, so we can proceed with the SNS issue.

I will need the log following the SNS reboot.

You can add tracing for SNS by editing:
/usr/local/simple_network_setup/rc.network
to change line 37 from blank to:
set -x #DEBUG

[LateAdopter]

Hello rerwin

The network cable is always connected and the router is always on.

If I boot Fatdog it always connects. In the messages I get the r8169 eth0 bit followed by the dhcpcd bit.

In BP64 I can run Netwiz and select eth0 - Auto DHCP and it connects the same way as Fatdog but, if I reboot, it does not reconnect. The messages then only contain the r8169 eth0 bit and no dhcpcd. But I can run dhcpcd myself and it will connect.

In BP64 I can run SNS but it does not detect the network at all, so I can't save the configuration and the question of SNS reboot does not arise.

I haven't played with the copy I did for your tests. So I can do any more tests you need starting from where the last post left off.

[rerwin]

OK, Thanks. Let's first work on the eth0 problem, which should be simpler.

In file /etc/rc.d/rc.network_eth, change line 66 from:
sleep 1.5 #190209
to:
sleep 5 #190209

That should make it wait up to 20 seconds instead of 6. It appears that fatdog waits up to 15 seconds, so this should work. (Be sure there is no /etc/simple_network_setup/connections file so we test the eth0 default.)


While you are making changes, go ahead and add tracing for SNS, but make /usr/local/simple_network_setup/rc.network line 37 to be:
set -x; exec >&2 #DEBUG

That will trace not only the reboot, but also when SNS first starts the interface. The log to get will be /tmp/xerrs.log for the initial setup of sns, while the reboot trace (once we get there) will be in /tmp/bootsysinit.log.

[LateAdopter]

Hello rerwin

I edited the files.
Then reboot
Then run SNS from Internet connection wizard.

Here are the two logs. The network connects at startup but SNS setup doesn't connect.

[rerwin]

Thank you.

So it takes between 10 and 15 seconds to detect the ethernet connection - solved.

I assume sns has the same issue, but let's confirm that. Edit:
/usr/local/simple_network_setup/sns

line 42 to change the blank line to:
set -x; exec >&2 #DEBUG

and change line 1001 from:
[ $i -lt 5 ] && sleep 1.5 #190209 190216

to:
[ $i -lt 5 ] && sleep 5 #190209 190216

I plan to change SNS and NetWiz to wait up to 15 seconds unless anyone thinks it should be more (20 seconds?). That period could slow reboot connection if the the PC has multiple ethernet cards and the first one checked is unplugged, so we don't want to make it any longer than necessary.

Has anyone failed to reconnect wired ethernet in Fatdog or the other dogs? Fatdog uses a 15-second timeout and I assume the others do also.

[LateAdopter]

Hello rerwin

I edited the file
Reboot & wait for network to connect
Run SNS setup from Internet connection wizard. - Success.

On another BP64 copy, I tested connection at boot with sleep 2.5 in rc.network_eth and that worked OK

In Netwiz selecting eth0 - Auto DHCP, it connects when the progress bar is one character before the C in Connecting.

[rerwin]

LateAdopter,

Thanks for the log and the added info about the "2.5" test.

That tells me that you need slightly over 10 seconds to get your connection. So, 15 seconds seems adequate.

I have now uploaded new network_connect_update, SNS and NetWiz packages (to the usual place in this thread) as my solution to the problems.

Please try all 3 of them (including the replaced NetWiz 2.1.2), with no network manager, SNS and NetWiz. You might add the "set -x; exec >&2" line to 'sns' to see how long the connection takes. The new code tries every second, so I expect to see it take 11 seconds.

Thank you very much for all of your help with this.
Richard

[LateAdopter]

Hello rerwin

I deleted the savefile and made a new one
Installed the three packages
Reboot - success
Added DEBUG to SNS
Run SNS from Internet Connection Wizard - success
Run Netwiz, test - success, Auto DHCP - success

Thank you too!

[rerwin]

Thanks for the log. It took 8 seconds (8 1-second sleeps) to connect. My previous conclusion was wrong; the time had to be between 5 and 10 seconds.

I have re-uploaded the packages to correct the timeout tests in order to realize the full 15 seconds instead of 14. So please do not save anywhere else the copies you used.

I think we are done! Good work.

[ozsouth]

@Rerwin - Sorry, I don't think I've understood the process here. My previous issue was that every 5th (or so) bootup sns wouldn't connect WIRELESSLY. OK in frisbee (I did a frisbee eth0 test for MikeW's benefit). I installed sns-2.3.1 in ScPup64 & couldn't connect wirelessly at all.

[rerwin]

ozsouth,

Thanks for reporting your wireless failure with SNS 2.3.1. Let's work the problem!

. We can start with your editing:
/usr/local/simple_network_setup/rc.network

line 38, to replace the blank line with:
set -x; exec >&2 #DEBUG

Do the same kind of thing in line 43 of:
/usr/local/simple_network_setup/sns

Then save and post or send me the file, /tmp/xerrs.log after the connection attempt fails.

That way I can probably see where things go wrong.

We can work the reconnect issue after we solve the initial problem.

Richard

[ozsouth]

Thanks @rerwin - I will do tests over the next day or so.

[B.K. Johnson]

Hello Richard
I must have missed some earlier posts and only now saw the post today. Wanting to know the version of SNS I was using,
I tried sns --version, then sns -help as you see below.

Code: Select all

# sns --version
OK
OK
dhcpcd[8096]: sending commands to master dhcpcd process
Selected interface 'wlan0'
OK
dhcpcd[8116]: sending signal ARLM to pid 21495
dhcpcd[8116]: waiting for pid 21495 to exit
eth0      no wireless extensions.

lo        no wireless extensions.

EXIT="EXITNOW"
SNS rc.network: waited for ethernet interfaces: seconds=30
#
# sns -help
EXIT="EXITNOW"
SNS rc.network: waited for ethernet interfaces: seconds=30
#

1. This was a surprising output to standard commands (or not so standard]. Perhaps a change will be in order?.
2. It took about 30 secs for each command to return to prompt. Each time the attached screenshot or similar appeared and sns was not usable after. It may resume working when I reboot. Can't do that ATM.