pixelserv pixelserv - A Better One-pixel Webserver for Adblock

[dugaduga]

Pixselserv keeps on crashing & shutting down. After updating to Diversion 4.1 & the latest entware packages. https://pastebin.com/PxbM4Dwj
the last part of the log you see i had restarted it... I'm not seeing anything in the logs suggestive of why this is taking place. somewhere between the beginning and the end of the log it was disabled. I would like to re-install pixelserv-tls and see if that does it, do you know the command?

 

[kvic]

I'm not seeing anything in the logs suggestive of why this is taking place.

For troubleshooting, turn on log to LEVEL 5 and let it run until suspected crash. Be prepared..as your syslog will be very busy with messages from pixelserv-tls.

I would like to re-install pixelserv-tls and see if that does it, do you know the command?

You'd better ask in Diversion thread for Diversion way. I believe you'll lose some customisation if you go with stock Entware operation.

 

[kvic]

Not entirely sure about that. It is possible /jffs got resized and parts of some files left out of bounds, yes? So restoring from backup maybe a good idea.

Resize in the context of expansion into unpartitioned free space should have old data retained. The issue is likely as mentioned a few posts back: one nvram variable, one directory, and/or two files are overwritten as part of FW upgrade. The fix is simply re-run the script to generate a new cert.

 

[Sebastien Bougie]

Hello,

Since i update Diversion to 4.1 pixelserv is not running. in the router log i'm getting

pixelserv-tls[30243]: Abort: Address already in use - :*:80

Any idea?

 

[Asad Ali]

Hello,

Since i update Diversion to 4.1 pixelserv is not running. in the router log i'm getting

pixelserv-tls[30243]: Abort: Address already in use - :*:80

Any idea?

In router web UI under LAN->DHCP Server->Basic Config

Check your IP Poll Starting Address and make sure it starts from one step ahead than your pixelserv-tls IP.

 

[Sebastien Bougie]

In router web UI under LAN->DHCP Server->Basic Config

Check your IP Poll Starting Address and make sure it starts from one step ahead than your pixelserv-tls IP.

Hello,

my pool is starting at .100 and my pixelsrv is .2 for sure this is not the problem it was working before

But thanks for the tips

 

[M@rco]

Since i update Diversion to 4.1 pixelserv is not running. in the router log i'm getting

pixelserv-tls[30243]: Abort: Address already in use - :*:80

If you installed or updated pixelserv-tls from outside Diversion, use Diversion to reinstall pixelserv-tls: https://www.snbforums.com/threads/diversion-the-router-adblocker.48538/page-19#post-430325

Otherwise, re-install Diversion: https://www.snbforums.com/threads/diversion-the-router-adblocker.48538/page-19#post-430479

It should solve your problem. It doesn't seem pixelserv-tls related (although the errors make it look that way).

 

[Sebastien Bougie]

If you installed or updated pixelserv-tls from outside Diversion, use Diversion to reinstall pixelserv-tls: https://www.snbforums.com/threads/diversion-the-router-adblocker.48538/page-19#post-430325

Otherwise, re-install Diversion: https://www.snbforums.com/threads/diversion-the-router-adblocker.48538/page-19#post-430479

It should solve your problem. It doesn't seem pixelserv-tls related (although the errors make it look that way).

Thanks for your help its working now

 

[kvic]

If you installed or updated pixelserv-tls from outside Diversion,

Just a minor correction. I believe the few users popped up here and complaint about pixelserv-tls "Abort: Address already in use - :*:80" did their update from Diversion. I would bet they even don't know how to update Entware/pixelserv-tls outside Diversion..

 

[Jack Yaz]

Just a minor correction. I believe the few users popped up here and complaint about pixelserv-tls "Abort: Address already in use - :*:80" did their update from Diversion. I would bet they even don't know how to update Entware/pixelserv-tls outside Diversion..

Initial install via diversion here, ran into the issue by running a manual opkg update/upgrade. A quick copy paste of a backed up S80pixelserv sorted it though, once I saw pixelserv was trying to listen on all available IPs and blocking my WebUI and Nginx from starting

 

[Adamm]

Initial install via diversion here, ran into the issue by running a manual opkg update/upgrade

Same thing happened here, but I was able to figure out pretty quickly what went wrong. Prehaps this is to be taken into consideration for future versions as part of another "self-heal" function.

 

[GDT]

I tried to configure certificate following this and imported ca.crt in my Windows PC but chrome gave me "NET::ERR_CERT_INVALID" when I try to open servstats page. From my android device I got a certificate error but I can reach the page.

Edit: I tried with Firefox and followed the linked procedure for it but I got "SEC_ERROR_BAD_SIGNATURE "

Edit2: I installed pixelserv-tls with Diversion and updated to the 2.2.0.rc2, I purged old certificate with Diversion and now I got no errors with Firefox. I did again the import wizard from Windows and now I can open servstats page from chrome but I got "NET::ERR_CERT_AUTHORITY_INVALID" and I need to make an exclusion for the website. The same happens with Edge or IE ("DLG_FLAGS_INVALID_CA")

Edit3: Ok, I fixed it by myself deleting everything and restarting from the beginning.

 

[kvic]

once I saw pixelserv was trying to listen on all available IPs and blocking my WebUI and Nginx from starting

That's the default configuration from Entware. Entware also has a few quirks in terms of upgrade process. If you install and upgrade manually a few times on packages which are services (such as pixelserv-tls), then you will have realised it and perhaps will have semi auto way to work around it. See this post for more info: https://www.snbforums.com/threads/diversion-the-router-adblocker.48538/page-20#post-430706

TLC replied subsequently and applied a fix. Perhaps worth a second look. Ping @thelonelycoder

 

[kvic]

Edit3: Ok, I fixed it by myself deleting everything and restarting from the beginning.

Between 2.1.3 beta and 2.2.0-rc.2, there was a fix and people shall purge old certs and only keep the CA cert (and its key).

The above info was mentioned in one of the post here, and on the release page on my blog. While old posts are hard to follow, the release page is quite easy. My users should read every word from me there.

But I understand people are busy, so let me add a few bold words to remind ppl there. Sometimes I'm also a good baby-sitter

 

[M@rco]

But I understand people are busy, so let me add a few bold words to remind ppl there. Sometimes I'm also a good baby-sitter

Thanks for the reminder Done

Two questions: Does pixelserv-tls need to be restarted after purging? Does it affect the stats?

 

[thelonelycoder]

Thanks for the reminder Done

Two questions: Does pixelserv-tls need to be restarted after purging? Does it affect the stats?

Diversion has a menu entry in ep to do that. It'll restart pixelserv-tls in the process, it's stats are therefore reset.

 

[thelonelycoder]

Same thing happened here, but I was able to figure out pretty quickly what went wrong. Prehaps this is to be taken into consideration for future versions as part of another "self-heal" function.

The new "self heal" function is to run Re-Install in the d menu of Diversion.

 

[Asad Ali]

Thanks for the reminder Done

Two questions: Does pixelserv-tls need to be restarted after purging? Does it affect the stats?

Yes and Yes.

 

[quant88]

The new beta (2.2.0-rc.2) for running fine for 4 days with options: -c 700
and encounter the following:

Sep 20 18:18:30 pixelserv-tls[29873]: Failed to create conn_handler thread. err: 11
Sep 20 18:18:30 pixelserv-tls[29873]: Failed to create conn_handler thread. err: 11
.
.
Sep 20 18:22:48 pixelserv-tls[29873]: Failed to create conn_handler thread. err: 11
Sep 20 18:23:23 pixelserv-tls[29873]: Out of memory. Cannot malloc receiver buffer.

With htop, I could tell there was still plenty of mem available. 200 MB of available 503 MB was used.
Not sure what has went wrong. Any idea ?
Meanwhile I have run with -c reduced to 350.

 

[kvic]

Are you running the version P update of the openssl lib?

Without the optimization flag, it's likely to cause such memory issue. Either switch back to version n-1c of the test lib (should be okay if you're not using any openssl related services such as SSH, OpenVPN etc from Entware).

Or wait a bit longer and see if I could offer you something better.