spdMerlin Multiple entries in GUI

[Ripshod]

Additional.
Logged into amtm and accessed spdMerlin and the GUI page gets mounted, as @alan6854321 saw.

Nov  2 13:07:47 ripshod dropbear[7299]: Child connection from 10.0.0.16:41973
Nov  2 13:07:47 ripshod dropbear[7299]: Pubkey auth succeeded for 'Gilly' with ssh-ed25519 key SHA256:K3p1xudvgwShuxGYqI23Vvnl*************** from 10.0.0.16:41973
Nov  2 13:08:01 ripshod spdMerlin: Mounting WebUI tab for spdMerlin
Nov  2 13:08:02 ripshod spdMerlin: Mounted spdMerlin WebUI page as user8.asp

I thought it best to leave it a few hours since the log I collected.

 

[Ripshod]

One other thing. The automatic speedtests for VPNs are being disabled again

 

[Martinski]

@Martinski how's this?

https://drive.google.com/file/d/16LFlVwaOXakAkBdJEJDW06r2EECSCF5R/view?usp=drive_link

I ran out of time to review the log. When I get some time during the week, I'll take a look.

BTW, when I tried to view the syslog, I get a page saying that I don't have access and that I need to request access.
Can you please store the syslog in PasteBin instead? Or perhaps on something like Pastes.io?

One other thing. The automatic speedtests for VPNs are being disabled again

I have not been able to duplicate this problem after the latest changes I made (see post #13). It's possible that whatever is causing the WebUI page not to get mounted on your system during reboot may also be intermittently causing other issues. Until I'm able to duplicate it, this issue remains elusive.

 

[Ripshod]

BTW, when I tried to view the syslog, I get a page saying that I don't have access and that I need to request access

Stupid permissions setting. Try the link again.

It's possible that whatever is causing the WebUI page not to get mounted on your system during reboot may also be intermittently causing other issues.

Agreed.

 

[Martinski]

Stupid permissions setting. Try the link again.

Thank you for providing the log file. The link did work. However, it looks like this is a curated or filtered log file because it's missing some key entries, like the call to the "/jffs/scripts/services-start" script. Also, I don't see the "ntpd: Started ntpd" entry, and some of the "/jffs/scripts/service-event" calls. Basically, much of the initial bootup sequence appears to be missing.

As mentioned before, I really need to see the *full* context of the reboot sequence, from start to finish, including when all existing add-ons are started. I now have an idea of where the problem might be, and I believe the clues are in the log, but I need the full context of what's happening to verify my assumptions.

There's a new 'develop' branch version 'v4.4.15_25110400' which generates more syslog messages along with process IDs.

When you get a chance, please update your current 'develop' version with the most recent one, and after reproducing the problem, please provide the log of the latest reboot sequence - the entire unfiltered sequence of events.

Thank you.

 

[Ripshod]

Thank you for providing the log file. The link did work. However, it looks like this is a curated or filtered log file because it's missing some key entries, like the call to the "/jffs/scripts/services-start" script. Also, I don't see the "ntpd: Started ntpd" entry, and some of the "/jffs/scripts/service-event" calls. Basically, much of the initial bootup sequence appears to be missing.

As mentioned before, I really need to see the *full* context of the reboot sequence, from start to finish, including when all existing add-ons are started. I now have an idea of where the problem might be, and I believe the clues are in the log, but I need the full context of what's happening to verify my assumptions.

There's a new 'develop' branch version 'v4.4.15_25110400' which generates more syslog messages along with process IDs.

View attachment 68739

When you get a chance, please update your current 'develop' version with the most recent one, and after reproducing the problem, please provide the log of the latest reboot sequence - the entire unfiltered sequence of events.

Thank you.

I never filtered anything out. That is the messages.log file as it was downloaded from the router, just as klogd stopped.
Without removing scribe and changing the log levels I don't see what else I can do right now (I'm at work)
I'll still try this latest build.

 

[Ripshod]

Okay @Martinski I've downloaded a full boot log (scribe removed), and DMed you a link due to the sensitive info it contains.

 

[Martinski]

Okay @Martinski I've downloaded a full boot log (scribe removed), and DMed you a link due to the sensitive info it contains.

Thank you for providing the log files covering the full reboot sequence of events. They gave me most of the details I needed to narrow down what I believe to be the root cause. Essentially, it was a race condition between 2 instances of the script being launched. One instance needs to initialize the internal structures for the VPN client interfaces (WireGuard and/or OpenVPN) as they come up during reboot. The other instance initializes the script to mount the WebUI page. This created a race condition for the locking mechanism.

There's a new 'develop' branch version 'v4.4.15_25110522' now available, which addresses the issue so that the WebUI page gets mounted *and* the VPN client interfaces are properly initialized as they come up during reboots.

When you get a chance, please update to the latest 'develop' version to test and verify the fix.

I'd like to express my full appreciation for your patience and willingness to help and collaborate with me while chasing down the root cause of the problems you found. I know it can be somewhat frustrating at times, so hats off to you for sticking with it for as long as you have.

 

[Ripshod]

Quick test of this new build this morning resulted in 5 out of 5 successful reboots, meaning the GUI mounted successfully with previously selected VPN slots all active. Great work.
I'll spend a little more time after work seeing if I can poke any holes in it,

 

[Ripshod]

Try as I might I can't make this build fall over. You're onto a winner @Martinski