pixelserv pixelserv - A Better One-pixel Webserver for Adblock

[Protik]

A question. Can I query the stats from command line? If yes, how can I do that?

 

[kvic]

A question. Can I query the stats from command line? If yes, how can I do that?

wget -qO - http://pixelserv-ip/servstats.txt

Alternatively you can install "lynx" from Entware to get the graphical version (/servstats) in ASCII graphic..

lynx -dump http://pixelserv-ip/servstats

 

[jrmwvu04]

A question. Can I query the stats from command line? If yes, how can I do that?

curl pixelserv.ip.address/servstats.txt - if you've changed ports or whatever else, adjust accordingly

 

[jrmwvu04]

Alternatively you can install "lynx" from Entware to get the graphical version (/servstats) in ASCII graphic..

Well, now I have to do that. Time to go back to 1995...

 

[jrmwvu04]

 

[kvic]

 

[kvic]

Installed and working perfectly. I am loving the prefetch ssl cache option. That would make everything much faster! The ssl cache restores the cache without any issues so far. Will let you know of any development.

The prefetch ssl cache shall jumpstart pixelserv-tls into better shape in shorter time. Without prefetch, it'll eventually get there still. Also note that, prefetch as in Km-test.4 only saves (on SIGTERM exit) part of the cache. Hence, frequent restart of pixelserv-tls is still not recommended.

 

[kvic]

Installed, will let you know.

I have faith in you crushing into the first bug

 

[kvic]

In Km-test.4, if you run pixelserv-tls with '-l 3', you'll see messages like below in your syslog:

On startup, prefetching certs:

Mar 10 00:19:01 Phaeo pixelserv-tls[21849]: pixelserv-tls 2.1.0-test.4 (compiled: Mar 10 2018 00:09:54) options: 192.168.1.3 -O 300 -A 344 -l 3
Mar 10 00:19:01 Phaeo pixelserv-tls[21849]: sslctx_tbl_load: _.moatads.com 252
Mar 10 00:19:01 Phaeo pixelserv-tls[21849]: sslctx_tbl_load: google-analytics.com 118
Mar 10 00:19:01 Phaeo pixelserv-tls[21849]: sslctx_tbl_load: _.adnxs.com 234
Mar 10 00:19:01 Phaeo pixelserv-tls[21849]: sslctx_tbl_load: _.go-mpulse.net 239
Mar 10 00:19:01 Phaeo pixelserv-tls[21849]: sslctx_tbl_load: _.google-analytics.com 274
Mar 10 00:19:01 Phaeo pixelserv-tls[21849]: sslctx_tbl_load: _.googlesyndication.com 216
Mar 10 00:19:01 Phaeo pixelserv-tls[21849]: sslctx_tbl_load: _.akamaihd.net 43
Mar 10 00:19:01 Phaeo pixelserv-tls[21849]: sslctx_tbl_load: _.servebom.com 183
Mar 10 00:19:01 Phaeo pixelserv-tls[21849]: sslctx_tbl_load: _.quantserve.com 5
Mar 10 00:19:01 Phaeo pixelserv-tls[21849]: sslctx_tbl_load: _.indexww.com 73
Mar 10 00:19:01 Phaeo pixelserv-tls[21849]: Listening on :192.168.1.3:344
Mar 10 00:19:01 Phaeo pixelserv-tls[21849]: Listening on :192.168.1.3:443
Mar 10 00:19:01 Phaeo pixelserv-tls[21849]: Listening on :192.168.1.3:80

When expired sessions are checked and flushed:

Mar  9 17:26:38 erx pixelserv-tls[12029]: sslctx_tbl_check_and_flush: expired TLS sessions (_.g.doubleclick.net) flushed.
Mar  9 17:26:38 erx pixelserv-tls[12029]: sslctx_tbl_check_and_flush: expired TLS sessions (192.168.10.1) flushed.
Mar  9 17:26:38 erx pixelserv-tls[12029]: sslctx_tbl_check_and_flush: expired TLS sessions (_.googletagservices.com) flushed.
Mar  9 17:26:38 erx pixelserv-tls[12029]: sslctx_tbl_check_and_flush: expired TLS sessions (_.scorecardresearch.com) flushed.

 

[jrmwvu04]

I have faith in you crushing into the first bug

I was hitting dailymail hard and it crashed. I've been stressing it since and not been able to replicate. But I had no such crashes on test 3.

uts    0d 00:10    process uptime
log    1    critical (0) error (1) warning (2) notice (3) info (4) debug (5)
kcc    1    number of active service threads
kmx    35    maximum number of service threads
kvg    3.68    average number of requests per service thread
krq    35    max number of requests by one service thread
req    1361    total # of requests (HTTP, HTTPS, success, failure etc)
avg    776 bytes    average size of requests
rmx    4864 bytes    largest size of request(s)
tav    23 ms    average processing time (per request)
tmx    9098 ms    longest processing time (per request)
slh    346    # of accepted HTTPS requests
slm    166    # of rejected HTTPS requests (missing certificate)
sle    0    # of rejected HTTPS requests (certificate available but bad)
slc    1    # of dropped HTTPS requests (client disconnect without sending any request)
slu    81    # of dropped HTTPS requests (unknown error)
sct    40    ssl cache: # of cached cert
sch    76    ssl cache: # of cache hit
scm    40    ssl cache: # of cache miss
scp    0    ssl cache: # of purge to free up slots

 

[vango44]

Any idea why I have these messages in my router log? (AC86U)

Mar  9 10:34:59 pixelserv-tls[19790]: Cannot use /opt/var/cache/pixelserv/_.sharethrough.com
Mar  9 10:34:59 pixelserv-tls[19790]: Cannot use /opt/var/cache/pixelserv/_.myfonts.net

Just those 2 files every few minutes.
Using 2.0.1

 

[kvic]

Any idea why I have these messages in my router log? (AC86U)

Mar  9 10:34:59 pixelserv-tls[19790]: Cannot use /opt/var/cache/pixelserv/_.sharethrough.com
Mar  9 10:34:59 pixelserv-tls[19790]: Cannot use /opt/var/cache/pixelserv/_.myfonts.net

Just those 2 files every few minutes.
Using 2.0.1

Pls run this in SSH

chown -R nobody /opt/var/cache/pixelserv

 

[kvic]

I was hitting dailymail hard and it crashed. I've been stressing it since and not been able to replicate. But I had no such crashes on test 3.

uts    0d 00:10    process uptime
log    1    critical (0) error (1) warning (2) notice (3) info (4) debug (5)
kcc    1    number of active service threads
kmx    35    maximum number of service threads
kvg    3.68    average number of requests per service thread
krq    35    max number of requests by one service thread
req    1361    total # of requests (HTTP, HTTPS, success, failure etc)
avg    776 bytes    average size of requests
rmx    4864 bytes    largest size of request(s)
tav    23 ms    average processing time (per request)
tmx    9098 ms    longest processing time (per request)
slh    346    # of accepted HTTPS requests
slm    166    # of rejected HTTPS requests (missing certificate)
sle    0    # of rejected HTTPS requests (certificate available but bad)
slc    1    # of dropped HTTPS requests (client disconnect without sending any request)
slu    81    # of dropped HTTPS requests (unknown error)
sct    40    ssl cache: # of cached cert
sch    76    ssl cache: # of cache hit
scm    40    ssl cache: # of cache miss
scp    0    ssl cache: # of purge to free up slots

I've been trying hard but not yet crashed on me.

I do experience some peculiar behaviour. Daily Mail after first few loads used to pop up instantly on both Chrome & Safari. Chrome remains butter smooth. On Safari somehow I can "feel" a bit slowness..

 

[vango44]

Pls run this in SSH

chown -R nobody /opt/var/cache/pixelserv

OK I think that fixed it.
Thanks!

 

[jrmwvu04]

I do experience some peculiar behaviour. Daily Mail after first few loads used to pop up instantly on both Chrome & Safari. Chrome remains butter smooth. On Safari somehow I can "feel" a bit slowness..

It is substantiated in my numbers. As I remarked earlier about the tav being exceptionally low for the dailymail torture test with Km test 3, those numbers are under test 4 reverted to builds prior to test 3. There is also a sluggishness as you describe. To be fair, I haven’t tested any other browsers to this point.

 

[Makaveli]

Thanks. Note that in both @Makaveli and your cases, near 90% (slc/req) of ad requests are HTTPS. But seems both of you don't have the CA cert imported in your main browsing devices (like desktop or phone). That's perfectly fine.

Though with the CA cert imported, you might notice a smoother browsing experience, especially on desktop with ad heavy sites such as CNN, Foxnews and Daily Mail..

Perhaps consider spending sometime over the weekend to go through this guide Import ca.crt into clients.

Thank you sir this is what I was looking for and was hoping you would post this. Saved me from searching the last 50+ pages.

Going update to test 4 and do the ca cert import.

 

[kvic]

It is substantiated in my numbers. As I remarked earlier about the tav being exceptionally low for the dailymail torture test with Km test 3, those numbers are under test 4 reverted to builds prior to test 3. There is also a sluggishness as you describe. To be fair, I haven’t tested any other browsers to this point.

I looked into the details and also repeated on Km-test.3. It's same between Km-test.3 and Km-test.4 for Safari.

Checked Safari console. Quite a lot of errors. So seems that 1) Daily Mail's backend changed a bit in the past few days and/or 2) my blocklist changed a bit in the past few days. Somehow Chrome is able to better cope with the change in DM's site. Way fewer errors in Chrome console.

Once the test pages are fully cached in browser and pixelserv-tls, Chrome performs the same on test.3 and test.4. So does Safari. Chrome is noticeable faster in both versions as of current Daily Mail (from the past 12 hours).

 

[jrmwvu04]

I looked into the details and also repeated on Km-test.3. It's same between Km-test.3 and Km-test.4 for Safari.

Checked Safari console. Quite a lot of errors. So seems that 1) Daily Mail's backend changed a bit in the past few days and/or 2) my blocklist changed a bit in the past few days. Somehow Chrome is able to better cope with the change in DM's site. Way fewer errors in Chrome console.

Once the test pages are fully cached in browser and pixelserv-tls, Chrome performs the same on test.3 and test.4. So does Safari. Chrome is noticeable faster in both versions as of current Daily Mail (from the past 12 hours).

I had pondered that earlier. This will always be an issue. It’s not going to be scientific to test because it will be, practically speaking, impossible to control the variables “blocked server requested” and/or “website reaction to pixelserv”. But so it goes.

The errors, I haven’t seen but have no difficulty attributing to safari handling things in a non standard way. That’s sort of the modus operandi at Apple, it seems.

At any rate, test 4 still up and stable after the one and only crash.

 

[kvic]

The errors, I haven’t seen but have no difficulty attributing to safari handling things in a non standard way. That’s sort of the modus operandi at Apple, it seems.

If you open Web Inspector, then select "Console" tab and load the DM page. What are there? For me, it's like this.. 46 errors.

 

[jrmwvu04]

If you open Web Inspector, then select "Console" tab and load the DM page. What are there? For me, it's like this.. 46 errors.

Likewise, like this.